- Kernel - mailweb.openeuler.org

[PATCH OLK-6.6 0/6] OLK-6.6 Bperf
by Xiaomeng Zhang 09 Jun '25

09 Jun '25

Ian Rogers (1): perf test bpf-counters: Add test for BPF event modifier Tengda Wu (2): perf stat: Support inherit events during fork() for bperf perf test: Use sqrtloop workload to test bperf event Veronika Molnarova (1): perf test stat_bpf_counter.sh: Stabilize the test results Xiaomeng Zhang (2): perf stat: Increase perf_attr_map entries perf stat: Fix incorrect display of bperf when event count is 0 tools/lib/perf/include/perf/bpf_perf.h | 1 + tools/perf/builtin-stat.c | 1 + tools/perf/tests/shell/stat_bpf_counters.sh | 79 ++++++++++----- tools/perf/util/bpf_counter.c | 61 +++++++++--- tools/perf/util/bpf_skel/bperf_follower.bpf.c | 98 +++++++++++++++++-- tools/perf/util/bpf_skel/bperf_u.h | 5 + tools/perf/util/target.h | 1 + 7 files changed, 198 insertions(+), 48 deletions(-) -- 2.34.1

2 7

[openeuler:OLK-5.10 2953/2953] kernel/locking/lockdep.o: warning: objtool: lockdep_unregister_key()+0x2ff: unreachable instruction
by kernel test robot 09 Jun '25

09 Jun '25

Hi Waiman, First bad commit (maybe != root cause): tree: https://gitee.com/openeuler/kernel.git OLK-5.10 head: dd32d51934d00fd319caec8710bd1a438bd69859 commit: b91170be2b1f1ef673f5e79c7a4beb5673a8f881 [2953/2953] locking/lockdep: Avoid potential access of invalid memory in lock_class config: x86_64-randconfig-123-20250609 (https://download.01.org/0day-ci/archive/20250609/202506092002.S7MmpM4X-lkp@…) compiler: clang version 20.1.2 (https://github.com/llvm/llvm-project 58df0ef89dd64126512e4ee27b4ac3fd8ddf6247) reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20250609/202506092002.S7MmpM4X-lkp@…) If you fix the issue in a separate patch/commit (i.e. not just a new version of the same patch/commit), kindly add following tags | Reported-by: kernel test robot <lkp(a)intel.com> | Closes: https://lore.kernel.org/oe-kbuild-all/202506092002.S7MmpM4X-lkp@intel.com/ All warnings (new ones prefixed by >>): >> kernel/locking/lockdep.o: warning: objtool: lockdep_unregister_key()+0x2ff: unreachable instruction -- 0-DAY CI Kernel Test Service https://github.com/intel/lkp-tests/wiki

1 0

[PATCH OLK-5.10 v6 00/11] arm64: Support xcall prefetch
by Jinjie Ruan 09 Jun '25

09 Jun '25

Support xcall prefetch. Changes in v6: - Only use one reserved kabi in task_struct for xcall. - Free prefetch item after eventpoll_release_file() to avoid possible competition between ep_prefetch_item_enqueue() and pfi free. - Fix "echo > /proc/xcall/cpu_list" show problem. Changes in v5: - Fix dentry free before free_prefetch_item() and then UAF in prefetch work by cancel_work_sync() before free pfi and move free_prefetch_item() to the first of __fput(). - Fix __queue_work() oops bug. - /proc/xcall/mask_list -> /proc/xcall/cpu_list - Fix xcall_prefetch print data type. - Update prefetch tracepoint, move xcall points definition to separete header file. - Make xcall_prefetch depends on EPOLL. - Adjust xcall prefetch buf to 16KB by default. - Add xcall_prefetch_init() function. - Only use write lock in free_prefetch_item(). - Simplify xcall_read() and add some comment. - Update commit message. Changes in v4: - Enable FAST_SYSCALL/IRQ and XCALL_PREFETCH by default. - Fix xint sleeping function called from invalid context bug. - free_prefetch_item() in file_free() instead of filp_close(), which fix the alloc-queue and free competition. - Fix kernel_read() warning for FMODE_READ not set. - Add sock_from_file() limit for prefetch. - Check NULL for rc_work. - Simplfy the cpumask interface code, and rename to "/proc/xcall/mask_list". - Remove the xcall_cache_pages_order interface. - tracepoint update: fd -> file. - Simplify the xcall_read() function again. - Handle copy_to_user() return value. - Remove unused XCALL_CACHE_QUEUED. - Update the commit message. Changes in v3: - Add XCALL_PREFETCH config to isolate feature code. - Split the /proc/xxx interface code out to independent patches, which will make it clear. - Update the cpumask interface to "/proc/xcall/numa_mask", and it can set the numa mask of all numas one time. - Add xcall select count to make xcall_cache_pages_order adjust safe. - Introduce xcall_read_start/end() to make it clear. - Simplify the xcall_read() function. - Use cpumask_next() instead of get_nth_cpu_in_cpumask() function. - Use independent cpu select policy function. - Remove some unnecessary pr_err(). - Update the commit message. Changes in v2: - Upadte the xcall prefetch state machine, remove the queued state and add prefetch, cancel states. - Remove the pfi lock and use atomic variables. - Change the 'xcall select' semantics and simplify the code a lot. - Remove keep_running, remove unrelated code. - Remove the count in struct read_cache_entry, and use percpu hit/miss count. - Remove sync mode, so remove struct read_cache_entry in struct task_struct. - Use hash table to find prefetch item for a file, which will not change the file struct KABI. - Use rwlock instead of spinlock for hash table. - Use alloc_page() instead kmalloc() to align 4KB. - Update the commit message. Jinjie Ruan (7): arm64: Introduce Xint software solution arm64: Add debugfs dir for xint eventpoll: xcall: Support async prefetch data in epoll xcall: Add /proc/xcall/prefetch dir for performance tuning xcall: Add /proc/xcall/cpu_list for performance tuning xcall: eventpoll: add tracepoint config: Enable FAST_SYSCALL/IRQ and XCALL_PREFETCH by default Liao Chen (1): revert kpti bypass Yipeng Zou (3): arm64: Introduce xcall a faster svc exception handling arm64: Faster SVC exception handler with xcall xcall: Introduce xcall_select to mark special syscall arch/Kconfig | 81 +++++ arch/arm64/Kconfig | 2 + arch/arm64/configs/openeuler_defconfig | 7 + arch/arm64/include/asm/cpucaps.h | 2 + arch/arm64/include/asm/exception.h | 3 + arch/arm64/kernel/asm-offsets.c | 3 + arch/arm64/kernel/cpufeature.c | 56 ++++ arch/arm64/kernel/entry-common.c | 22 ++ arch/arm64/kernel/entry.S | 183 ++++++++++- arch/arm64/kernel/syscall.c | 57 ++++ drivers/irqchip/irq-gic-v3.c | 130 ++++++++ fs/eventpoll.c | 409 +++++++++++++++++++++++++ fs/file_table.c | 1 + fs/proc/base.c | 150 +++++++++ fs/read_write.c | 10 +- include/linux/fs.h | 35 +++ include/linux/hardirq.h | 5 + include/linux/irqchip/arm-gic-v3.h | 13 + include/linux/sched.h | 4 + include/linux/xcall.h | 19 ++ include/trace/events/xcall.h | 106 +++++++ kernel/fork.c | 25 ++ kernel/irq/debugfs.c | 33 ++ kernel/irq/internals.h | 18 ++ kernel/irq/irqdesc.c | 19 ++ kernel/irq/proc.c | 10 + kernel/softirq.c | 73 +++++ 27 files changed, 1473 insertions(+), 3 deletions(-) create mode 100644 include/linux/xcall.h create mode 100644 include/trace/events/xcall.h -- 2.34.1

2 12

[PATCH OLK-6.6 0/6] OLK-6.6 Bperf
by Xiaomeng Zhang 09 Jun '25

09 Jun '25

Ian Rogers (1): perf test bpf-counters: Add test for BPF event modifier Tengda Wu (2): perf stat: Support inherit events during fork() for bperf perf test: Use sqrtloop workload to test bperf event Veronika Molnarova (1): perf test stat_bpf_counter.sh: Stabilize the test results Xiaomeng Zhang (2): perf stat: Increase perf_attr_map entries perf stat: Fix incorrect display of bperf when event count is 0 tools/lib/perf/include/perf/bpf_perf.h | 1 + tools/perf/builtin-stat.c | 1 + tools/perf/tests/shell/stat_bpf_counters.sh | 79 ++++++++++----- tools/perf/util/bpf_counter.c | 61 +++++++++--- tools/perf/util/bpf_skel/bperf_follower.bpf.c | 98 +++++++++++++++++-- tools/perf/util/bpf_skel/bperf_u.h | 5 + tools/perf/util/target.h | 1 + 7 files changed, 198 insertions(+), 48 deletions(-) -- 2.34.1

2 7

[PATCH OLK-5.10] IMA: use real_inode to get the i_version
by Gu Bowen 09 Jun '25

09 Jun '25

hulk inclusion category: bugfix bugzilla: https://gitee.com/openeuler/kernel/issues/ICDKL2 -------------------------------- overlayfs has two types of inode, the overlayfs inode generated by overlayfs and the real inode of the file. When IMA does the measurement, process_measurement() will try to detect file content changes for files on a overlayfs filesystem based on the i_version number of the real inode. But now comparing with value of overlayfs inode, results in always re-evaluating the file's integrity. Therefore, ima_collect_measurement() should update iint->iversion with real_inode iversion. Also, ima_check_last_writer() should compare i_version base on real_inode. This patch is based on the implementation of upstream patch (see below Link tag). Due to merging the pre-patch to resolve conflicts introduces KABI changes, we don't fix this with the mainline version. Link: https://web.git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/comm… Fixes: b836c4d29f27 ("ima: detect changes to the backing overlay file") Signed-off-by: Gu Bowen <gubowen5(a)huawei.com> --- security/integrity/ima/ima_api.c | 2 +- security/integrity/ima/ima_main.c | 7 +++++-- 2 files changed, 6 insertions(+), 3 deletions(-) diff --git a/security/integrity/ima/ima_api.c b/security/integrity/ima/ima_api.c index d88d7fb9f9a5..69bb71331e3f 100644 --- a/security/integrity/ima/ima_api.c +++ b/security/integrity/ima/ima_api.c @@ -278,7 +278,7 @@ int ima_collect_measurement(struct integrity_iint_cache *iint, * which do not support i_version, support is limited to an initial * measurement/appraisal/audit. */ - i_version = inode_query_iversion(inode); + i_version = inode_query_iversion(real_inode); hash.hdr.algo = algo; /* Initialize hash digest to 0's in case of failure */ diff --git a/security/integrity/ima/ima_main.c b/security/integrity/ima/ima_main.c index c211a2c5f297..efa20bd18b97 100644 --- a/security/integrity/ima/ima_main.c +++ b/security/integrity/ima/ima_main.c @@ -222,16 +222,19 @@ static void ima_check_last_writer(struct integrity_iint_cache *iint, { fmode_t mode = file->f_mode; bool update; + struct inode *real_inode; if (!(mode & FMODE_WRITE)) return; mutex_lock(&iint->mutex); if (atomic_read(&inode->i_writecount) == 1) { + real_inode = d_real_inode(file_dentry(file)); + update = test_and_clear_bit(IMA_UPDATE_XATTR, &iint->atomic_flags); - if (!IS_I_VERSION(inode) || - !inode_eq_iversion(inode, iint->version) || + if (!IS_I_VERSION(real_inode) || + !inode_eq_iversion(real_inode, iint->version) || (iint->flags & IMA_NEW_FILE)) { iint->flags &= ~(IMA_DONE_MASK | IMA_NEW_FILE); iint->measured_pcrs = 0; -- 2.25.1

2 1

[PATCH OLK-6.6] arm64/mpam: Add MPAM manual
by Zeng Heng 09 Jun '25

09 Jun '25

hulk inclusion category: feature bugzilla: https://gitee.com/openeuler/kernel/issues/IC88OM -------------------------------- Add user manual for the MPAM features. Signed-off-by: Zeng Heng <zengheng4(a)huawei.com> --- Documentation/arch/arm64/mpam.md | 587 +++++++++++++++++++++++++++++++ 1 file changed, 587 insertions(+) create mode 100644 Documentation/arch/arm64/mpam.md diff --git a/Documentation/arch/arm64/mpam.md b/Documentation/arch/arm64/mpam.md new file mode 100644 index 000000000000..60f62cd7f781 --- /dev/null +++ b/Documentation/arch/arm64/mpam.md @@ -0,0 +1,587 @@ +MPAM 用户手册 +========= + +# 1 MPAM 简介 + +MPAM（Memory system component Partioning and Monitoring）是Arm Architecture v8.4的拓展特性。用于解决服务器系统中，混合部署不同类型业务时，由于共享资源的竞争（L3/L2 Cache，MATA），而带来的关键应用性能下降或者系统整体性能下降问题。 + +MPAM的应用可针对不同业务，将同时作用于硬件访存路径上产生的竞争和冲突进行隔离控制，从而帮助提升服务器利用率，降低服务部署成本。 + +本手册只适用于OLK-6.6软件版本。 + +# 2 内核编译选项 + +配置 CONFIG_ARM64_MPAM=y 后，即使能MPAM完整功能。 + +# 3 内核启动参数 + +启动流程默认关闭MPAM，启动MPAM初始化需要**cmdline**添加**arm64.mpam**参数配置后重启机器。 + +# 3 接口总览 + +MPAM功能通过resctrl文件系统呈现，挂载点位于 */sys/fs/resctrl* 。系统启动后，需要手动挂载resctrl文件系统。 + +## 3.1 系统挂载参数 + +resctrl可以通过添加挂载参数支持多种挂载方式，具体指令如下： + +~~~ +# mount -t resctrl resctrl [-o cdp[,cdpl2][,debug][,l2]] /sys/fs/resctrl +~~~ + +挂载参数包括： + +* cdp: 针对L3缓存，根据访问指令和访问数据分别配置。 +* cdpl2: 针对L2缓存，根据访问指令和访问数据分别配置。 +* debug: 使能调试接口访问。 +* l2: 使能L2缓存配置和监控功能，默认关闭MPAM L2功能。 + +## 3.2 resctrl 系统目录介绍 + +### 3.2.1 Info 目录 + +Info 目录包含有关已启用资源的信息，每个资源都有其自己的子目录，子目录的名称反映了资源的名称。 + +每个子目录包含以下与分配相关的文件： + +缓存资源（L3/L2）子目录包含以下与分配相关的文件： + +**num_closids**: 适用于该资源的有效CLOSID（Class of Service ID）数量。内核会以所有已启用资源中最小的CLOSID数量作为限制。 + +**cbm_mask**: 适用于该资源的有效位掩码（bitmask）。 + +**min_cbm_bits**: 写入掩码时必须设置的连续位的最小数量。 + +**shareable_bits**: 与其他执行实体共享资源的位掩码。用户在设置独占缓存分区时可以使用此字段。 + +**bit_usage**: 标注了资源所有实例的使用情况的容量位掩码。说明如下： + + * *0*: 对应区域未使用。当系统的资源已被分配，且在“bit_usage”中发现“0”时，这表明资源被浪费了。 + + * *H*: 对应区域仅由硬件使用，但可供软件使用。如果资源的“shareable_bits”中有位被设置，但这些位并未全部出现在资源组的分配方案中，则“shareable_bits”中出现但未分配给资源组的位将被标记为“H”。 + + * *X*: 对应区域可供共享，并被硬件和软件使用。这些位同时出现在“shareable_bits”和资源组的分配中。 + + * *S*: 对应区域由软件使用，并可供共享。 + + * *E*: 对应区域被一个资源组独占使用，不允许共享。 + +**sparse_masks**: 指示是否支持CBM（Capacity Bit Mask）中的非连续1值。 + + * *0*: 仅支持CBM中的连续1值。 + + * *1*: 支持CBM中的非连续1值。 + +MB（Memory bandwidth，内存带宽）子目录包含以下与分配相关的文件： + +**min_bandwidth**: 用户可以请求的最小内存带宽百分比。 + +**bandwidth_gran**: 内存带宽百分比分配的粒度。分配的带宽百分比会四舍五入到硬件上可用的下一个控制步长。可用的带宽控制步长为： +~~~ +min_bandwidth + N * bandwidth_gran +~~~ + +**delay_linear**: 指示延迟刻度是线性还是非线性的。该字段仅用于信息参考。 + +如果支持监控功能，则会存在一个名为 L3_MON 和 MB_MON 的目录，其中包含以下文件： + +**num_rmids**: 可用的RMID（Resource Monitoring ID）数量。这是可以创建的“CTRL_MON”+“MON”组的最大数量。 + +**mon_features**: 如果为该资源启用了监控功能，则列出监控事件。例如： +~~~ +# grep . /sys/fs/resctrl/info/*_MON/mon_features +/sys/fs/resctrl/info/L3_MON/mon_features:llc_occupancy +/sys/fs/resctrl/info/MB_MON/mon_features:mbm_total_bytes +~~~ + +**max_threshold_occupancy**: 读/写文件，提供一个最大值（以字节为单位），低于此设定值下，之前使用过的LLC_occupancy计数器可以被考虑重新分配使用。 + +> 请注意，一旦释放了RMID（资源监控ID），它可能不会立即可用，因为RMID仍然与之前使用该RMID的缓存行相关联。因此，这些RMID会被放入一个“待定”列表中，并在缓存占用量降低后再次检查。如果系统中存在大量处于“待定”状态的RMID，但它们尚未准备好被使用，用户在执行mkdir操作时可能会看到-EBUSY错误。 + +> max_threshold_occupancy是一个用户可配置的值，用于确定在什么占用量下可以释放一个RMID。 + +最后，在 Info 目录的顶层有一个名为 **last_cmd_status** 的文件。每次通过文件系统发出“命令”（例如创建新目录或写入任何控制文件）时，该文件都会被重置。如果命令成功，文件内容将显示为“ok”。如果命令失败，它将提供比文件操作错误返回更详细的信息。例如： +~~~ +# echo "MB:1=110" > schemata +-bash: echo: write error: Invalid argument +# cat /sys/fs/resctrl/info/last_cmd_status +MB value 110 out of range [0,100] +~~~ + +### 3.2.2 资源分配和监控 + +资源组在resctrl文件系统中以目录的形式表示。默认组是根目录，刚挂载后，它拥有系统中的所有任务和CPU，并且可以充分利用所有资源。 + +在支持RDT（资源分配技术）控制功能的系统中，可以在根目录下创建额外的目录，这些目录指定了每种资源的不同数量（参见下面的“schemata”）。根目录和这些额外的顶级目录在下文中被称为“CTRL_MON”组。 + +在支持RDT监控的系统中，根目录和其他顶级目录中包含一个名为“mon_groups”的目录，在其中可以创建额外的目录来监控其父“CTRL_MON”组中任务的子集。这些在本文档的其余部分被称为“MON”组。 + +删除一个目录会将其所代表的组拥有的所有任务和CPU移动到其父目录。删除一个创建的“CTRL_MON”组将自动删除其下所有的“MON”组。 + +支持将“MON”组目录移动到一个新的父“CTRL_MON”组，以便在不影响其监控数据或分配的任务的情况下更改“MON”组的资源分配。此操作不适用于监控CPU的“MON”组。目前，除了简单地重命名“CTRL_MON”或“MON”组之外，不支持其他任何移动操作。 + +所有组包含以下文件： + +**tasks**: 读取此文件将显示属于该组的所有任务列表。将任务ID写入该文件会将任务添加到该组中。可以通过用逗号分隔任务ID来添加多个任务。任务将按顺序分配。在尝试分配任务时遇到的任何单个失败都会导致操作中止，而在失败之前已添加到组中的任务将保留在组中。失败信息将记录到/sys/fs/resctrl/info/last_cmd_status。 + +如果该组是一个“CTRL_MON”组，则任务将从之前拥有该任务的“CTRL_MON”组中移除，同时也会从任何拥有该任务的“MON”组中移除。如果该组是一个“MON”组，则任务必须已经属于该组的“CTRL_MON”父组。任务将从任何之前的“MON”组中移除。 + +**cpus**: 读取此文件将显示该组拥有的逻辑CPU的位掩码。将掩码写入该文件将向该组添加或移除CPU。与“tasks”文件类似，维护了一个层级结构，其中“MON”组只能包含其父“CTRL_MON”组拥有的CPU。 + +**cpus_list**: 与“cpus”类似，但使用CPU范围而不是位掩码。 + +当启用控制功能时，所有“CTRL_MON”组还将包含以下文件： + +**schemata**: 列出该组可用的所有资源。每种资源都有自己的行和格式——详细信息请参见下文。 + +**size**: 类似于“schemata”文件的显示，但显示的是每种资源分配的字节大小，而不是表示分配的位。 + +**mode**: 资源组的“mode”决定了其分配的共享方式。“shareable”资源组允许共享其分配，而“exclusive”资源组则不允许。 + +**ctrl_hw_id**: 仅在启用调试选项时可用。硬件用于控制组的标识符。在arm64架构上，即是 PARTID。 + +当启用监控功能时，所有“MON”组还将包含以下文件： + +**mon_data**: 包含一组按L3域和MB域事件组织的文件。例如，在具有两个L3域的系统中，将存在子目录“mon_L3_00”和“mon_L3_01”。 + +每个子目录中都有一个文件对应每个事件（例如“llc_occupancy”、“mbm_total_bytes”）。在“MON”组中，这些文件提供了组中所有任务的当前事件值。在“CTRL_MON”组中，这些文件提供了“CTRL_MON”组中所有任务以及所有“MON”组中任务的总和。有关使用方法的更多详细信息，请参见示例部分。 + +**mon_hw_id**: 仅在启用调试选项时可用。硬件用于监控组的标识符。 + +以下为resctrl文件系统目录树： + +~~~ +/sys/fs/resctrl(根分组) + ├── cpus # bitmask方式显示根分组关联的vcpu + ├── cpus_list # cpu list方式显示根分组关联的vcpu + ├── ctrl_hw_id # 硬件用于控制组的标识符 + ├── info # 用于显示属性信息及错误提示信息 + │ ├── L3 + │ │ ├── bit_usage # 标注了资源所有实例的使用情况的容量位掩码 + │ │ ├── cbm_mask # 系统所支持的最大cache way bitmask，一个bit代表一个cache way + │ │ ├── min_cbm_bits # 使用schemata所能配置的最小cache way bitmask + │ │ ├── num_closids # L3能够提供创建控制组的最大数量 + │ │ ├── shareable_bits # 当前所有cbm_mask全部shareable，支持后续扩展 + │ │ └── sparse_masks # 指示是否支持CBM（Capacity Bit Mask）中的非连续1值 + │ ├── L3_MON + │ │ ├── max_threshold_occupancy # 低于此设定值下，之前使用过的LLC_occupancy计数器可以被考虑重新分配使用 + │ │ ├── mon_features # 列出监控事件 + │ │ └── num_rmids # 可创建控制组和监控组的总数 + │ ├── last_cmd_status # 操作错误提示 + │ ├── MB + │ │ ├── bandwidth_gran # 带宽百分比配置粒度 + │ │ ├── delay_linear # 指示延迟刻度是线性还是非线性的 + │ │ ├── min_bandwidth # 最小带宽配置百分比 + │ │ └── num_closids # 同L3 num_closid + │ └── MB_MON + │ ├── mon_features # 同L3 mon_features + │ └── num_rmids # 同L3 num_rmids + ├── mode # 资源组的 mode 决定了其分配的共享方式 + ├── mon_data + │ ├── mon_L3_01 # 标号代表L3 cache id + │ │ └── llc_occupancy # 表示当前分组所关联的pid/vcpu在该区域上实际占用L3 Cache大小，下同 + │ ├── mon_L3_122 + │ │ └── llc_occupancy + │ ├── mon_MB_00 # 标号代表numa id + │ │ └── mbm_total_bytes # 表示当前分组所关联的pid/vcpu在该区域上内存带宽流量大小，下同 + │ └── mon_MB_01 + │ └── mbm_total_bytes + ├── mon_groups # 创建监控组目录 + ├── mon_hw_id # 硬件用于监控组的标识符 + ├── schemata # 资源使用配置接口 + ├── size # 显示的是每种资源分配的字节大小 + └── tasks # 显示与根组关联的pid +~~~ + +### 3.2.3 控制组配置接口 Schemata 文件 + +**schemata**文件中的每一行描述一种资源。每行以资源的名称开头，后面跟着该资源在系统中每个实例上要应用的具体值。 + +#### Cache IDs + +在当前一代的系统中，每个插槽（socket）有一个L3缓存，而L2缓存通常仅由一个核心上的超线程共享，但这并不是架构上的强制要求。我们可能会在一个插槽上有多个独立的L3缓存，或者多个核心共享一个L2缓存。因此，我们不使用“插槽”或“核心”来定义共享资源的逻辑CPU集合，而是使用“Cache ID”（缓存ID）。 + +在给定的缓存级别上，这将在整个系统中是一个唯一的数字（但不能保证是一个连续的序列，可能会有间隔）。要查找每个逻辑CPU的ID，请查看 /sys/devices/system/cpu/cpu*/cache/index*/id。 + +#### Cache Bit Masks（CBM，缓存位掩码） + +对于缓存资源，我们使用位掩码来描述可用于分配的缓存部分。掩码的最大值由每种CPU型号定义（并且可能因不同的缓存级别而异）。该值在resctrl文件系统的“info”目录中提供，位于info/{resource}/cbm_mask。 + +#### L3 缓存配置 + +当未启用CDP（代码/数据缓存划分）时，L3 schemata的格式为： + +~~~ +L3:<cache_id0>=<cbm>;<cache_id1>=<cbm>;... +~~~ + +当启用CDP时，L3控制被拆分为两个独立的资源，因此您可以分别为代码和数据指定独立的掩码，如下所示： + +~~~ +L3DATA:<cache_id0>=<cbm>;<cache_id1>=<cbm>;... +L3CODE:<cache_id0>=<cbm>;<cache_id1>=<cbm>;... +~~~ + +读取schemata文件将显示所有域上所有资源的状态。在写入时，您只需要指定您希望更改的值。 + +例如使用默认方式挂载，设置L3缓存位掩码只有4位： + +~~~ +# mount -t resctrl resctrl /sys/fs/resctrl/ +# cat /sys/fs/resctrl/schemata +L3:1=fffffff;122=fffffff + +# echo "L3:122=3c0;" > /sys/fs/resctrl/schemata +# cat /sys/fs/resctrl/schemata +L3:1=fffffff;122=00003c0 +~~~ + +使用开启CDP方式挂载，设置L3 data缓存位掩码只有4位： + +~~~ +# mount -t resctrl resctrl /sys/fs/resctrl/ -o cdp +# cat /sys/fs/resctrl/schemata +L3DATA:1=fffffff;122=fffffff +L3CODE:1=fffffff;122=fffffff + +# echo "L3DATA:122=3c0;" > schemata +# cat /sys/fs/resctrl/schemata +L3DATA:1=fffffff;122=00003c0 +L3CODE:1=fffffff;122=fffffff +~~~ + +#### L2 缓存配置 + +L2 缓存配置功能默认关闭，需要通过显式添加 l2 挂载参数，才会使能L2缓存配置功能。使能L2功能以后，系统关闭 cpuidle powerdown 功能和cpu下线功能。 + +L2 schemata的格式是： + +~~~ +L2:<cache_id0>=<cbm>;<cache_id1>=<cbm>;... +~~~ + +使用“cdpl2”挂载选项可以在L2上支持CDP： + +~~~ +L2DATA:<cache_id0>=<cbm>;<cache_id1>=<cbm>;... +L2CODE:<cache_id0>=<cbm>;<cache_id1>=<cbm>;... +~~~ + +L2 缓存配置示例，设置L2缓存位掩码只有4位： + +~~~ +# mount -t resctrl resctrl /sys/fs/resctrl/ -o l2 +# cat schemata +L2:4=000ff;8=000ff + +# echo "L2:4=f;" > schemata +# cat schemata +L2:4=0000f;8=000ff +~~~ + +使用“cdpl2”挂载选项： + +~~~ +# mount -t resctrl resctrl /sys/fs/resctrl/ -o l2cdp,l2 +# cat schemata +L2DATA:4=000ff;8=000ff +L2CODE:4=000ff;8=000ff + +# echo "L2DATA:4=f;" > schemata +# cat schemata +L2DATA:4=0000f;8=000ff # 控制组 L2 DATA 只能使用L2 cache4的4个cache way +L2CODE:4=000ff;8=000ff +~~~ + +#### MB 内存带宽分配 + +对于内存带宽资源，默认情况下，用户通过指定总内存带宽的百分比来控制该资源。 + +每种CPU型号的最小带宽百分比值是预定义的，可以通过info/MB/min_bandwidth查询。分配的带宽粒度也取决于CPU型号，可以在info/MB/bandwidth_gran中查询。可用的带宽控制步长为：min_bw + N * bw_gran。中间值会被四舍五入到硬件上可用的下一个控制步长。 + +MB schemata的格式是： + +~~~ +MB:<cache_id0>=bandwidth0;<cache_id1>=bandwidth1;... +~~~ + +MB 带宽配置示例： + +~~~ +# cat /sys/fs/resctrl/schemata +MB:0=0000100;1=0000100 + +# echo "MB:0=50" > /sys/fs/resctrl/schemata +# cat /sys/fs/resctrl/schemata +MB:0=0000050;1=0000100 # 降低控制组 MB 内存带宽使用上限为50% +~~~ + +### 3.2.4 资源分配规则 + +当一个任务正在运行时，以下规则定义了它可用的资源： +1. 如果任务属于一个非默认组，则使用该组的分配方案（schemata）。 +2. 否则，如果任务属于默认组，但运行在一个被分配给某个特定组的CPU上，则使用该CPU所属组的分配方案。 +3. 否则，使用默认组的分配方案。 + +### 3.2.5 监控组配置方法 + +读取控制组和监控组的监控数据，可通过 mon_data 目录接口读取监控数据： + +~~~ +# grep . mon_data/*/* +mon_data/mon_L3_01/llc_occupancy:73276416 +mon_data/mon_L3_122/llc_occupancy:11875328 +mon_data/mon_MB_00/mbm_total_bytes:32806 +mon_data/mon_MB_01/mbm_total_bytes:31700 +~~~ + +其中，mon_data读取监控数据文件分别： + * llc_occupancy 代表 L3缓存当前占用量，单位 Byte + * mbm_total_bytes 代表内存带宽瞬时流量，单位 MB/s + +支持在控制组下创建子监控组，监控父控制组监控对象的子集： + +~~~ +# cd /sys/fs/resctrl/p1 +# cd mon_groups/ && mkdir m1 # 监控组只能监控，m1分组资源配置跟随p1分组 +# echo '0-1' > cpus_list +# grep . mon_data/mon_*/* +mon_data/mon_L3_01/llc_occupancy:18432 +mon_data/mon_L3_122/llc_occupancy:1024 +mon_data/mon_MB_00/mbm_total_bytes:0 +mon_data/mon_MB_01/mbm_total_bytes:0 +~~~ + +控制组监控的是控制组本身及所有子监控组的监控值之和。 + +## 3.3 QoS 增强特性 + +### 3.3.1 PRI 优先级设置 + +对共享资源优先级进行配置，包括 L3PRI 和 MBPRI： + +~~~ +# cat schemata + MBPRI:0=0000007;1=0000007 + L3PRI:1=0000003;122=0000003 + +# echo "MBPRI:0=0000003" > schemata +# cat schemata + MBPRI:0=0000003;1=0000007 # 降低控制组 MB numa0的优先级 + L3PRI:1=0000003;122=0000003 + +# echo "MBPRI:0=0000003" > schemata + +~~~ + +优先级设置数字越大，即优先级越高，反之，数字越小，优先级越低。 + +> MBPRI 默认值为 3，MBPRI合法值范围 [0,7]。L3PRI 默认值为 0，L3PRI合法值范围 [0,3]。 + +### 3.3.2 MIN 限低值设置 + +共享资源实际使用占比低于设置值，会自动提高对该资源使用优先级，包括 L3MIN 和 MBMIN： + +~~~ +# cat schemata + MBMIN:0=00100 + L3MIN:1=00100;5=00100 + +# echo "MBMIN:0=00050" > schemata +# cat schemata + MBMIN:0=00050 + L3MIN:1=00100;5=00100 + +# echo "L3MIN:1=00050" > schemata +# cat schemata + MBMIN:0=00050 + L3MIN:1=00050;5=00100 +~~~ + +L3MIN 和 MBMIN 接口接受的输入参数为百分比，即设置值与总资源（内存带宽/缓存占用量）的占比。 + +> MBMIN 和 L3MIN 默认值为 0，合法值范围都为 [0,100]。 + +### 3.3.3 HDL 强制隔离设置 + +当MBHDL=1，限制MB共享资源使用量不能超出MB设置值，若MBHDL=0，则允许空闲情况下，MB共享资源使用量超过MB设置值： + +~~~ +# cat schemata + MBHDL:0=0000001;1=0000001 + +# echo "MBHDL:0=0000000" > schemata +# cat schemata + MBHDL:0=0000000;1=0000001 # 关闭MB在numa0上的强制限制功能 +~~~ + +> MBHDL 默认值为 1，合法值范围 [0,1]。 + +### 3.3.4 MAX 资源上限设置 + +设置允许分配的缓存容量的最大百分比，包括 L3MAX 接口： + +~~~ +# cat schemata + L3MAX:1=00100;5=00100 + +# echo "L3MAX:1=00050" > schemata +# cat schemata + L3MAX:1=00050;5=00100 # 降低L3分配的缓存最大容量百分比 +~~~ + +> MB 和 L3MAX 默认值为 100，MB合法值范围 [1,100]，L3MAX合法值范围 [0,100]。 + +## 3.4 外设 IO 流量管控 + +### 3.4.1 控制组绑定外设 + +MPAM 提供通过绑定 iommu_group ID，对设备IO流量进行带宽限制和监控。 + +譬如，控制网卡设备 eno2 ，首先查找该设备 PCI_SLOT 信息： + +~~~ +# cat /sys/class/net/eno2/device/uevent | grep PCI_SLOT +PCI_SLOT_NAME=0000:35:00.1 +~~~ + +或者通过 ethtool 工具查看 bus-info 信息： + +~~~ +# ethtool -i eno2 | grep bus-info +bus-info: 0000:35:00.1 +~~~ + +按照设备总线信息，查找到该设备所属的 iommu_group ： + +~~~ +# find /sys/kernel/iommu_groups/ -name "0000:35:00.1" +/sys/kernel/iommu_groups/17/devices/0000:35:00.1 +~~~ + +或者通过 lspci 工具查看 iommu_group 信息： + +~~~ +# lspci -vvv -s 0000:35:00.1 | grep "IOMMU group" + IOMMU group: 17 +~~~ + +将查询到的group id 17，通过 tasks 接口绑定到指定的控制组下： + +~~~ +# cd /sys/fs/resctrl/p1/ +# echo "iommu_group:17" > tasks +# cat tasks +iommu_group:17 # 此时iommu_group 17 已被绑定到控制组p1 +~~~ + +### 3.4.2 查看外设带宽流量 + +控制组绑定外设所属的iommu组后，可以查看设备流量带宽： + +~~~ +# grep . mon_data/mon_MB_0*/* +mon_data/mon_MB_00/mbm_total_bytes:0 +mon_data/mon_MB_01/mbm_total_bytes:4230 +~~~ + +### 3.4.3 配置外设带宽流量 + +通过MB配置接口，可以实现限制设备的流量带宽： + +~~~ +# echo "MB:1=0000001" > schemata +# cat schemata + MB:0=0000100;1=0000000 + +# grep . mon_data/*/* +mon_data/mon_MB_00/mbm_total_bytes:0 +mon_data/mon_MB_01/mbm_total_bytes:1208 +~~~ + +# 4 控制组和监控组配置使用示例 + +/sys/fs/resctrl 默认为根分组，根分组可以创建若干个控制组，一个控制组既可以关联一组pid/tid，也可以关联一组cpu集合。 + +创建一个新的控制组，关联 pid/tid： + +~~~ +# cd /sys/fs/resctrl/ && mkdir p1 +# cd p1 && echo $$ > tasks # 关联当前shell进程pid到p1组 +# cat tasks # 可查看成功关联的pid +29190 +29607 +~~~ + +也可以选择关联 cpu： + +~~~ +# cd p1 && echo '0-1' > cpus_list +# cat cpus_list # 可查看关联的cpu +0-1 +~~~ + +查看可创建的控制组和监控组数量： + +~~~ +# cat info/L3/num_closids # 可在info对应的资源的目录下查看closid数量，即可以创建控制组的数量 +32 +# cat info/MB/num_closids +32 +# cat info/L3_MON/num_rmids # 可以创建控制组和监控组的总数量 +128 +# cat info/MB_MON/num_rmids +128 +~~~ + +通过配置控制组可达到隔离 L3 Cache/Memory Bandwidth 效果，通过读取对应分组mon_data接口可以获取该组资源占用情况，比如对一个控制组限制L3 Cache使用： + +~~~ +# cat info/L3/cbm_mask # 查看info目录下对应资源的属性 +fffffff +# cat info/L3/min_cbm_bits +1 + +# cd /sys/fs/resctrl/p1 +# cat schemata + MB:0=0000100;1=0000100 + L3:1=fffffff;122=fffffff + +# echo 'L3:0=1' > schemata # 配置1条cache way给p1分组 +# cat schemata + MB:0=0000100;1=0000100 # 若此时该组关联pid/cpu，那么该pid/cpu产生的访存请求只会分配到这条cache + L3:1=0000001;122=fffffff +~~~ + +对控制组限制MB使用： + +~~~ +# cat info/MB/min_bandwidth # 和配置L3类似，也可以查看MB的相关信息 +1 +# cat info/MB/bandwidth_gran +1 # 可知，配置带宽最小百分比是1%，颗粒度是1% +# cat schemata + MB:0=0000100;1=0000100 + L3:1=0000001;122=fffffff +# echo 'MB:0=1' > schemata +# cat schemata + MB:0=0000001;1=0000100 + L3:1=0000001;122=fffffff +~~~ + +支持在控制组下创建子监控组： + +~~~ +# cd /sys/fs/resctrl/p1 +# cd mon_groups/ && mkdir m1 # 监控组只能监控，m1分组资源配置跟随p1分组 +# ls m1/ +cpus cpus_list mon_data tasks +# echo '0-1' > cpus_list +# cat cpus_list +0-1 +# grep . mon_data/mon_*/* +mon_data/mon_L3_01/llc_occupancy:18432 +mon_data/mon_L3_122/llc_occupancy:1024 +mon_data/mon_MB_00/mbm_total_bytes:0 +mon_data/mon_MB_01/mbm_total_bytes:0 +~~~ -- 2.25.1

2 1

[PATCH OLK-6.6 0/6] OLK-6.6 Bperf
by Xiaomeng Zhang 09 Jun '25

09 Jun '25

Ian Rogers (1): perf test bpf-counters: Add test for BPF event modifier Tengda Wu (2): perf stat: Support inherit events during fork() for bperf perf test: Use sqrtloop workload to test bperf event Veronika Molnarova (1): perf test stat_bpf_counter.sh: Stabilize the test results Xiaomeng Zhang (2): perf stat: Increase perf_attr_map entries perf stat: Fix incorrect display of bperf when event count is 0 tools/lib/perf/include/perf/bpf_perf.h | 1 + tools/perf/builtin-stat.c | 1 + tools/perf/tests/shell/stat_bpf_counters.sh | 79 ++++++++++----- tools/perf/util/bpf_counter.c | 61 +++++++++--- tools/perf/util/bpf_skel/bperf_follower.bpf.c | 98 +++++++++++++++++-- tools/perf/util/bpf_skel/bperf_u.h | 5 + tools/perf/util/target.h | 1 + 7 files changed, 198 insertions(+), 48 deletions(-) -- 2.34.1

2 7

[PATCH OLK-5.10] add xcall epoll prefetch selftest
by Yuntao Liu 09 Jun '25

09 Jun '25

--- .../testing/selftests/xcall_prefetch/Makefile | 37 ++++ .../selftests/xcall_prefetch/echo_client.c | 119 +++++++++++++ .../xcall_prefetch/echo_client_multi | Bin 0 -> 21896 bytes .../selftests/xcall_prefetch/echo_server.c | 110 ++++++++++++ .../selftests/xcall_prefetch/echo_test.c | 52 ++++++ .../xcall_prefetch/mulit_close_test.c | 48 +++++ .../xcall_prefetch/multi_write_test.c | 77 ++++++++ .../xcall_prefetch/signal_recovery_test.c | 164 ++++++++++++++++++ .../xcall_prefetch/thundering_herd_test.c | 119 +++++++++++++ .../testing/selftests/xcall_prefetch/utils.h | 54 ++++++ .../xcall_prefetch/xcall_prefetch_test.sh | 39 +++++ 11 files changed, 819 insertions(+) create mode 100644 tools/testing/selftests/xcall_prefetch/Makefile create mode 100644 tools/testing/selftests/xcall_prefetch/echo_client.c create mode 100755 tools/testing/selftests/xcall_prefetch/echo_client_multi create mode 100644 tools/testing/selftests/xcall_prefetch/echo_server.c create mode 100644 tools/testing/selftests/xcall_prefetch/echo_test.c create mode 100644 tools/testing/selftests/xcall_prefetch/mulit_close_test.c create mode 100644 tools/testing/selftests/xcall_prefetch/multi_write_test.c create mode 100644 tools/testing/selftests/xcall_prefetch/signal_recovery_test.c create mode 100644 tools/testing/selftests/xcall_prefetch/thundering_herd_test.c create mode 100644 tools/testing/selftests/xcall_prefetch/utils.h create mode 100755 tools/testing/selftests/xcall_prefetch/xcall_prefetch_test.sh diff --git a/tools/testing/selftests/xcall_prefetch/Makefile b/tools/testing/selftests/xcall_prefetch/Makefile new file mode 100644 index 000000000000..5bd509df66e0 --- /dev/null +++ b/tools/testing/selftests/xcall_prefetch/Makefile @@ -0,0 +1,37 @@ +CFLAGS += -Wall -Wextra -g +TEST_GEN_FILES := echo_server echo_client echo_test multi_write_test thundering_herd_test mulit_close_test signal_recovery_test + +# enable debug +# echo_client: CFLAGS += -DDEBUG=1 +# echo_server: CFLAGS += -DDEBUG=1 +# multi_write_test: CFLAGS += -DDEBUG=1 +# thundering_herd_test: CFLAGS += -DDEBUG=1 +# mulit_close_test: CFLAGS += -DDEBUG=1 +# signal_recovery_test: CFLAGS += -DDEBUG=1 + +all: $(TEST_GEN_FILES) + +echo_server: echo_server.c + $(CC) $(CFLAGS) $< -o $@ + +echo_client: echo_client.c + $(CC) $(CFLAGS) $< -o $@ + +echo_test: echo_test.c + $(CC) $(CFLAGS) $< -o $@ + +multi_write_test: multi_write_test.c + $(CC) $(CFLAGS) $< -o $@ + +thundering_herd_test: thundering_herd_test.c + $(CC) $(CFLAGS) $< -o $@ + +mulit_close_test: mulit_close_test.c + $(CC) $(CFLAGS) $< -o $@ + +signal_recovery_test: signal_recovery_test.c + $(CC) $(CFLAGS) $< -o $@ + +.PHONY: all + +include ../lib.mk diff --git a/tools/testing/selftests/xcall_prefetch/echo_client.c b/tools/testing/selftests/xcall_prefetch/echo_client.c new file mode 100644 index 000000000000..2e3c41b2b14e --- /dev/null +++ b/tools/testing/selftests/xcall_prefetch/echo_client.c @@ -0,0 +1,119 @@ +#define _GNU_SOURCE +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <sys/socket.h> +#include <netinet/in.h> +#include <unistd.h> +#include <sys/epoll.h> +#include <arpa/inet.h> +#include <pthread.h> +#include <errno.h> + +#include "./utils.h" + +#define DEBUG + +void *client(void *arg) +{ + int thread_id = *(int *)arg; + int sock = 0; + struct sockaddr_in serv_addr; + struct epoll_event ev, events[MAX_EVENTS]; + const char* test_messages[MAX_MSGS]; + + if ((sock = socket(AF_INET, SOCK_STREAM, 0)) < 0) { + handle_error("client:%d create socket failed\n", thread_id); + } + serv_addr.sin_family = AF_INET; + serv_addr.sin_port = htons(PORT); + + if (inet_pton(AF_INET, "127.0.0.1", &serv_addr.sin_addr) <= 0) { + close(sock); + handle_error("client:%d invalid address\n", thread_id); + } + if (connect(sock, (struct sockaddr *)&serv_addr, sizeof(serv_addr))) { + close(sock); + handle_error("client:%d connection failed\n", thread_id); + } + + debug_printf("Thread %d connected\n", thread_id); + + int epoll_fd = epoll_create1(0); + if (epoll_fd == -1) { + handle_error("client:%d epoll_create1 failed\n", thread_id); + } + + ev.events = EPOLLIN; + ev.data.fd = sock; + if (epoll_ctl(epoll_fd, EPOLL_CTL_ADD, sock, &ev) == -1) { + handle_error("client:%d epoll_ctl add sock failed\n", thread_id); + } + + test_messages[0] = "hello world."; + test_messages[1] = generate_number_string(100); + test_messages[2] = generate_number_string(1024); + test_messages[3] = generate_number_string(4096); + + + for (int i = 0; i < MAX_MSGS; i++) { + const char* msg = test_messages[i]; + ssize_t msg_len = strlen(msg); + if (send(sock, msg, msg_len, 0) != msg_len) { + handle_error("client:%d send failed\n", thread_id); + } + + debug_printf("Client:%d send %s\n", thread_id, msg); + + int nfds = epoll_wait(epoll_fd, events, MAX_EVENTS, -1); + if (nfds == -1) { + handle_error("client:%d epoll_wait failed\n", thread_id); + } + + for (int n = 0; n < nfds; ++n) { + if (events[n].data.fd == sock) { + char buffer[BUFFER_SIZE] = {0}; + ssize_t read_bytes = read(sock, buffer, BUFFER_SIZE); + + if (read_bytes <= 0) { + close(sock); + if (read_bytes == 0) { + handle_error("client:%d server disconnected\n", thread_id); + } else { + handle_error("client:%d read failed\n", thread_id); + } + } else { + debug_printf("Client:%d received %s\n", thread_id, buffer); + if (strncmp(msg, buffer, msg_len) != 0) { + handle_error("client:%d error: response does not match sent message\n", thread_id); + } + } + } + } + } + + close(sock); + pthread_exit(NULL); +} + +int main() +{ + pthread_t threads[NUM_THREADS]; + int ret; + + for (int i = 0; i < NUM_THREADS; i++) { + ret = pthread_create(&threads[i], NULL, client, &i); + if (ret != 0) { + perror("pthread_create failed"); + continue; + } + } + + for (int i = 0; i < NUM_THREADS; i++) { + if (threads[i] != 0) { + pthread_join(threads[i], NULL); + } + } + + return 0; +} \ No newline at end of file diff --git a/tools/testing/selftests/xcall_prefetch/echo_client_multi b/tools/testing/selftests/xcall_prefetch/echo_client_multi new file mode 100755 index 0000000000000000000000000000000000000000..36aa08d6e12f6e8ec73b2a7f8eb567225eb0b2f8 GIT binary patch literal 21896 zcmeHPYjhmNm98GCr5TMLBTKd=Ti9cR9ShLQFKmpBY-{is%MWY|>@4hfq-jY5qZwr$ zevlX&WF?*mWo;G$$@&mO&LIiO$uT*B@G{7}Y-bbJ>=JMun>7jKNWo-{fJ8uG_Ph0H zwK|&JJvsYpqta2|d%wE(R@JSV?&_}U-sfw+#-V9Ur5v_`k#ug7KpJJn!Dh*TG_obE z2**p<#VikUF2_WEqaer?>8xWcbxJ%3l<X>q8HZL0m?^4iNR;fdC68O=D9SuSCOe&2 zDcfmLfzVS_<(uVsMLx69a#?>xt#%ZfiW*^8v3x8Yk#S+tUfB^$cJrj&JZY!s1*#B| zqAH#g6Z&5w{n=?f86hcZx9gO4cG@J&n4(k$r7dqILw`+qo21>qEaAsa)i6_3m3JNN zC@%loq`G;tEH7I>49b2`REC$;MZ&F%7uH2WHIZ;MwYz5blEpQP7uF_Xwexuus9s+D z)0kSbp@qTC0AZ>-9s7l>{83)H%qRWs(jWZ%*E_C1J?FB63+}pchWni>DsCnnsyEq? z4khxZh6>5VPs2ayh<Mx%PGCe|F<$N>nk}tBhWu%dQHf{4D^QSkjHVE7&7yyI7W}0w z_|Yu*KV`vh$%608f(NtU7iGa0W@+yYS@fyD#*63dEcioN@ZK!=BU$i@EcoIq`06a} zs?MVScouw9798#wwfQenAkelu7-$bigOTv92J23C#*JVoVC)Vj8F8#D7>UH%%#533 z;V5g1MWaSrk_7_z8MXxy$zVJg=n5hy5o^1}NV0I$NCvu-u_!aTW06RpEg4~5Mppt^ ziDW$5*453rQ^^Eti^LM6g|?x<q(K@H1Fno{h;_n8LOR(MM4h^gcsv$YZn;4N>VYb> z-4bZ)yd}^c3`ZDMi%tfM*_!6Ht5yf**Df`=1vo`0bfL-3uU%yF3l_4#+O6vYAtP>d zgcC_4zIFZTNGxh>4Yo!Ml-AJ|i^^69gzC782(|-f2o3)^IQ!F){&f5cjCj(ZKMh<S z=ctb~<h%~5&Xbkla1otR-jG<lGVG$W+uKqu=lQwpPf|`zqkMQp$CHmA9{ws$M-Em* z0|7~0@6>f>piJOic|BA3eu+C9MZUt7eD-RA)45K%=D;}%uFkoXF=)ZjATwRC;8RSL zu@MVS=T(*JC7ut!dj4Hv!RdUhQlkau^9-5cx8T$Ul{Q&$Dnq4h7F@-G@^)Bo1l~-Y z7F<qYcwV;!ms28+@3!EDrWeNcTkv8Fe$axKSnxv@T#X^hzR!Y}TJ(=vaO?T<+ZJ37 z1#WfBg41<VrN=G!WCel_Sn#P9{FDW!bEZmXEO?~?C+I{3CL%Bqfxn>$yjL>gBY)3% zr@v4Cc^zZ^-hreebH?BEg7bMEh0LOlY8e~O%=s9kWF{m1PRbb`dLxs`9O5!vn1_d6 zF=e`74iB9$Wx7BQ4?SYabm1Hxdcc(Ff;l{N*p%r)IXrZ`DbodVc<6JcOc%!Cp`a<# z1#x)jI#Z?#;qcH(Q>F{x@KA#((}iz%XqG9{L}7TS)RgH0I6S18GF|wFhu-~EwU;h< zXum1bg%0gEWxBwj{iaM8HniWA>4Jv#D;d%4;?ZqlJid_csiA0(%KbLE*Cy|^$tj!M zWs?n?e3MOXvB~Rgvd<<jx5<laa*a)%W0Nnk$&+nzkxkZZ@+X&%jn@ZOx#69)=~w>V zpZ<-%=gkY7w)z?d8czHBmtTQlmYMu-(0sqWwq)i3bP9d|{mVZD;_ugoi7XvSPR38; zG2-#laAnSw%-m0U&r1sZ(_DYib402?^`|fRPyFF(|A~<tzxI;<l~0r9aL_9qI5X$k zdA}-q`u*%*PT4GVS&P4Cc`LF0^jpb7f8X+U2!lWWB$F8oVKBa=-vn8^3ASeWWIwbE znZ&ff&VT#)0q)_rAE-b5=r&I7*}+MFCrER;x0{px-5>`BlHQVP-~N*7^@mETTaK1g z-*l{`+88LQ?mAOaojO-iz4yYi^a~%{fdOk4({tXN-g~}h?|HUw3WD%7eq)0hQB%5c zWJCAj(C0hfGobnJ^QTV^zJ(lr+Bf3w**n5gWia>-8O%@yZ--C%-Y3hqoBCSb_qRRg z4-a&-x3~LG40=zDIMT13_;hCVh4hK^sT1cM+9_=y{Yo{2et&(R59(+9eekWFN&onu zr^olc25_10{1<+RDvV~ngt$WO^z&vLZa+`=F8-dvOBlk|h|`cief-q#GntUTFL%y7 zREO%^FyQa&CA5#Fd(VQN?EMvvR9pYa-V6U}vLlM^zlU9rk$B=$gwCIS{XPG8{ppVm zI{ZD)X-|^rGZfNi=}bB3M-<ZBXU}o&c~4~@FDU&I0`(xyW<02am56`)$zIRfnT&%y zduS$1i~bBK-8%`8zn|y$dj@j->F50Yi_$O@1+0Nq@4&vfeY_BKM;-u65n;24$EcsA zqx1=(|0JF6{OOlr7v+wUqmGV$JjT&7(@_^3JqtTF_*vMbd(R<aJ;&*oNm*NWxoNQs z7AJdi(GrS`tZJ{5x2&JwzI!M*ecTm~9xAjR<@bz?b@?GK2O?mDC>p<HdZz17_j=*u z*$`Ff3TkD#cOC)s)O2vJ0O~n_<_=C4z!Cr_d(X(;9(n^c>*FFub>D#($s_y>A|FG9 z6Vz$@M;yJt(R&1Wlrlw|o}j2et(DZ4_ELV*8j*-u{c2wQM;lR!X!R3_Ugk+^0_-1z zeY*EenD-o~<Jcf9H@*B8Av)RnDm19zp<Lbu66u$bKI7L0@Wb)rXXV!h{WVYH+SJFE z@_|y)dj=L54m8G})DV|v$OZkLAN>&a{r!|72MJ}1L2{b%(!C>8M9*;|Y~Om>pC<?V zC|AO%>6StGFs|yN+Zs^gx3^iczHeqtp{(WpJ`q+?0W<V3Q_Y?lp|L(Dq6ba$*AcTG z9t*9HlVv=BC3pTFKdm0#vs%w_>hGMAuS^1FR{K@B72~-84KU9jJ?Fb^=ZcfQyD3O- z<5Y9EzrW=uM#_nEIsX2u9R7YF=@b6`$)86ncqj+of!nn2NYCCQELnsfdpW8)I0*0% z%yITi`|c)Y#C6-TlfL`7<4gN|_tEfcdhsY-lBSM|X79WX2R**~w4S~9xo$hdgX-^F zKN9j^R>;q4Nqz8zKV>qfF~y^n@cExP<D)_*(nJI%A}|qwi3t3Ej{vO-P%4O~x>}96 z7i%NoXonZ;BOS?3Z#cRW%Qhi<u2{11(n^Ryvk|1WNZ5!bubdO|ibWQ$Sg4UY1*~EI z6}9#FYp`fa@7`c26gLuy0!x-$01C&VrU!dDa^Xj;(3o1I^)MHP$Um=+snls;$;rDb z7LSB#Errs8kf~#LM2kaarnS8ZBfb-X3xyN1(niSI5?bRL*N(3<UFk*3yRigictbHG z;f=<U-mYM>t<#IjfJIn>9Y%q<IwpfV8hLGc&a!5N2J;5C;}4n4e$WBT%DxSH26O<_ z`)(%F3F;coWX^(igVN2`!uK+na!dzzVlGfmlWrtT|9H}F-ORMz9&LJIzVi_B2&cUm z$8cjamo%F4G2vs_0|GzzKl@H5vx9sTd#)+2xVFT-%ekLjJ?+X%7hEzMD7U{6)Pp*4 zLjsY=Zxa7b;G1L}2(Q5Zo%pXt-8%^I6??wuSY1SqhNvFYz9jNq05$U*PocKL_7va> zv~yLiB_Ax=zKQ=8f6Qb^XVmstZcETanux$e1STRd5rK&aOhjNJ0uvGV+lzqOucY=T zsi`ULnW9uEh`E8PNXc>&k=j$D_CZaT`Iw&ar1l!A{ZKS7rKIvd_%stE`Iy|?r1li; zpb0RNB<!rr$0Kf@Zju#MdwyshP06}fhGc2RDy}AJbdS@Bw5NGFCABw6?Ww7twN?wh zO;%Xddw}QSUPDq<{tPKAJv;&B?L%=K!n0N`kH`Y?JX_^+OWbyF`5Bpyhp=2$?U3&O z&xqP@ZFg8NZ(bTD-6ZJ_NxLQ8FX<skk4kz>(g8`&NP14v3z9OqdCDWHSJHY(8ztQ& z=?+P|CAHW4f0CVzLSxP9)mM7wwX~+9$&|ODzIIV<eN97(iw(EUpI=+Quv)_Ga#)Cm z2clC?%l%u6)Y!C@IUQ<rH-;Y92GH+?m*P+dWIRKMGye#jJ96_M0LgJWG-n&hj{LWY zXn6)m?!EAkGj}TODRemW7eR7qaV=+FRTGGwtB0dWXc2Vj6uM1yIQ2&`WO6s+kW(`q ztH><3jhs9;{Q&qp=XXd+%ZtE!KJj_ZW29q1CwF}rd4h}Y!5S63m#T084bZ1^PtE}F z{9G!hpdzmpCa!77)m;0*I$T{;rt5QL<SGEKyB<Prp6kot3koW6q~C)3XHWpmrG=Wj zD&CO7xlr`v|IpbAtmsu}=0B#Be^&Hcs_2gjI;aGee<esUReKkh5^7x$T0p3WkkVh1 z=^fzmoOJZuN#HR!&!<S{IqB$mfWS}5@Ao)JN6%vft|Idj9HgV?pK&PsI@QhjfwKY@ zMHi^<f0A|kCe)oHCGUbbipZ<;!}8Z4<Q9Dey3RkB?S!e5<-f)~($RAfGA4Z=hbm{T z=G+UW>;}@!(}d4554p+LgwwJIsW_)r`~&h@Mm^<pX_Nns+?M@@qymvw!Sd(9O`emE zo;z?Ty%yQ|{T!sDhdQ(LT6oWYkb`vee2+jHUHL!aARRp?2wXv$Kj9!9J+BeiN8lSA zq@#z1YUx)|xBNeHkd7WY_2%)O^td_uC0Kc;aP|nXvpD+*v6pl9C1S7OECty^-H^wF z<5|zyy9nLJ*>4fs#@VNdjdJ!CV)t<Ncf=mx?8n6Z1K8ZJp#)b&*=Dpf_h}qlmE{KU zUn726`EKH`gQ9DCc^~l&;GHujQP>L~gl(>Krjxq4=p<Y^XO;XIJWwLe&rf;?q}=2D zcl|PS%;Z~1<v$8Ie*<it=iJK(|2-+c;a&%x6@MK9p?nONuwrsSIb6<Xyu+!Rxtzla zAAym_`Aa8_sG@1)^*63hU;|gweP2|n`OMh`Ui?Vi7w6^Ry6XBDXlSlp)XU-8fMc$! z3mzI>x52@FH@`A#Zhp0KxVMwfT-OO?Ib9!v_qh3#z+3PzY)CHn3JO5-7PNB88Vh#v zny~$B%3QK@H<!>}gQ?xgICF1sErZ(B_7HP#%>Nv*sRi?yds86=cWP?!Lgv1%@ELND zTY`goa}L$H@F~=@(7n}}gi!P(vfbBn3syuO?7qQ6CFHV-yI|umbaK2{&Ph&zr=%VE z7&?`+QEx{_3Gq3H5AB;4#~yk|r;99d4&S+N`gPKbw;bRvpg1ke95)vWl>_@&)x%O} zDfvaB{J6|NxDTE1Gk|Vf7kGcU*hPD&(9(Qr`ulRK!JpJY){X>mjh_*WcR-_@G;$6< z-cq>$jH66e>r_kSGl;UIT*^Obsl1is$rW@K$vM3L`pRX9iDOCu`N}zbE-=m5$sAKF znt3bVZ>gk^Ii^;{dG@jED|5)aqVx_fAG*G>l;q2#`!g+7v*iTn6Vc1Og;$k`2n@7T zHOTB=sO+u;s=s{lqrCcOTX=jND@vc@^0}7E&r<omBH`fGma5-L2XmBzD05spNo2o( z;^jQ*4D2aY&cng6p;Gu8Y~dh3^HYmv(wQG;nqq1$HHnA4jTKXM?ox6NyXf<fwb}XK z1gV$Z@B(;B+L2)>dY?djwAr(%B96m1R??MTt9Fxt<8WtHS{jM+dB1CQyx$#%LyXP( zBU-6d<*5sN2h1pGN4zg%lxkHoMSLE=Z8oG*ZBB8SJEwShaj7;b--CcSE@5u>3P9+? z60jK2Q&>*%mF{KkrS7$uZs<}0)yg9U!t&=bBDj9!6t8d-CiFFcu6Z!8f}aA;U_iP# zgD@4ALEr5j)h#Nr6tx64#hSLZjJ%Z0LJ4(1JXi3dO3R>m6}otm=|f-YCe1SG1LE2; z@=-ojvz9U0S_WfNrrPQ-75UE5=2p~dS67a<pTg8w@zqt8R8-|cOLZL>F4F**wggOf zBgB*PNx4d}j9p~)_?gPJWn+BKz?&pvT+HMxN3^`iG5dj>&l>B<bc45o_|W|+`a6eB zq0yv$TwH{sMkAf%T2_}z#OuP*wn!>u)ZtCLz~Y7SwY!dJs;)JR7kR8IV}v@q2}ExL zjRpV0Nn%d|JM5C-NTRlrN;TR#V*#<o6X;4slHuAm8kxyG-3Do(wuxjY90Q9XEMHo( zX5uv@TDFp^{G}wphohi4d60QDS(Czn5OT=10^~&ie1!@u;>mQ4xg6?ul2fJWy5pvD zU3*Y>zNNc5bk9@zr4Q)S+jZybx+_Iu@h)9Y>N7Ay&Z*NYi8wBuq?Zu0O~|M8GLmZb z*+gn|yb6@7U8}plsF$D83s>mcx?D%jjkS*1lXTZxZ|Ma`^-{;;DZ0~nNOv99r#TNC z)+_GUT|0rirn}aib{<fiUhe#o?tG8bN}RpA=Nr1~DUr8bp8}xbb$!bHdPQpVSh7pc zyI-HYU7zmk&`aOci&A=_6XY;=Rp_{+jQB~gb=|Me+^&0^!+Pn{`dnv^UhcTKT%Y_6 zz2dZ9`3>h$=Yx6$EGyUP7o$CnoNnhGx{F$t)aP&@?)2(;0?>2joH~slKB*TEQncpQ z=_Oa`g^p^J*72178Pc!Pr;_p^5uU5`;vJ+}UZ)q{uNQ3>I?j*^USUMnwmYU^_~HZ~ zpts1deuv>B-xmu7lfgie`8I8AZoXmjhK-xo^LN|=t$UJ2g2`9ku%1dofrd{YoQPS3 zXfWO_jMr~jBbbdFd|Uh*w<?#bR&How9Y)lM<3+ZBShK~8Vzg$97wPbRTp&OrZoz!i zA`y-*4zx#NyTZ|SltL1fDODO4lS*eSp2RAztn%6oqMXf5{+3O`0*jj6Mu4guj^fp~ zu5e_J(BT?*v5!Nlv~IYAe<~^^n~Rt0#`&q40l>?5CLiwJO_dYXY*-9Fk-$5HDxjO2 zctFLEmTlR%dR?Gt&E}QsM;)Y8`<I9ciT3i|@NHY&+_Giu^*$z+w&?|1@9Kst<jt(N zrYo6>8q0ZT!)*|Q@wU$8;;mnAO^3IpJ($=NZR?E3qOnw>CKW|gLp4cyUoU}7{(4|d zcRZHFYm2dX4Z0+TWptI%76~RgRW>bZ*R;2_TV&bSEn8)$i8!aCiEu~Mz><1rFfJOr zxoQ1qh?WSmO`F&9-nrF?#~9us#OggAOl#`i(HKV;%du+g#0oP@h#0lU;#hVTWo+2E zVO2At%M!r=4QM$^#`hPB2%{~&>o%=iz0S8)h^w17`nLI2TYeUJWs?_$o@q_Bw;S=% zIvIG4Q;gfi3y~dd4<)GYM^Bfs?JI8(4MrinPd7L5p93EWvT(DVyXP<BLE$5EG$<4c z-tdHUwRtmDjI?OTsPQ~nHw48Tl9-ow2BV>f5fJZ);x~=AQpFOzdWRJAcwV0^TQ~bG zLzC5Y#=4BUNPYd{h4YszTv8W{+_I>lzAhPyMG|%72*2REgb`^cK2f*34R6H;y5mN> zfz1G-t^wcmz71Qq;JwxOPSI=XWm<K{S-H+fzgzR<5*2l6TD{tW26<?pL)W{ki7w8$ z>|MTR)T%_!^)%$%iOV7-FF#_Y13I2QJ@<29D(WExOY(vzywQd)khps8s!Y~`(eUQG zr%@nOKT0(7)YufJo}+RQhcDot{14dtALROz&G%E4{<px4SAQ}e52uN@hTRLEi<O5_ zI-dS%u0NI8pDX?=H!EWUvqtTSdkg;>c0s7;vnuyvFq4#Eq4b{qc=+YOy%s&i>X?vt z8Ro2w3K7(w#m{!&9_Xw1E2|x(?<mD|Qz3%3XYoTfe&f~i?kxC2S@0ia!GD|u|5X-z zI18SG_)vT8@$vAujcqGQGl7>nDw%qYsx&SqoLcM^2xf*nEdxGY9InfP-<$=fM`RuZ zzoJn(mc+&NJ?3{@=1QENeUZMsAL)tQB;?p168s~_$NDDe;Vk$I(vSUi@|Rik{~-14 zPXV@P(f=R|?!x(q;%{%)WZ>SvtUjv)K3@FSWWnjt?(A{=+)DZmtOj|;X71;HJm#}) zb?e)o#XsB97Ed<R#&ER>wuS@AU<U(`Na9k7JCo551q6~^xNV6=F`WzqLa{(cB-R>? z1VWf};8LF2jZY_Zbw`Y(5vr{pUxU8I5Do<6@!*~SW&rU$tUVs=G6JDgSJxhxSR_p3 z!pSkJ<IUv)fom`e^96hxngaM*LDTjPE7z}G4NsV21$=&)>u=i30&AK#u3Fg~*m%t~ zTYOssTUV}X_QAsbk%vb0^#pt;0TU_unu7_^XB7UsZzs@KAI2&hd=A4~P2Mp58=G(Y zw1b(sP<$K0l&NIeAvgvQ-<%i&p?4C-7{uI(XrlCq2-+BE0=!v#wk1CMfS`>121Fo~ zhy}##1MS}kE!<uPQVHz8G{1#0#+#ZO(1$k0WQfh1V<3!!5K3%KBxLFQRzg&p_C1=) zRBz>2Y(*T+TA1y=M0`=hRHUZLIlSC3IVLL*z_&2eSfY<)jQiyaRSEi9h-r@DB10`U zg_;>l(Kzc@NX!f#b{gbkCw5}$IYx`sCiZk;x(}L+3)-nTOeziBxYc6wrcsM07PYt! zK%<j;EKqAIjBg3yK1yh<T(uTA4;_rFcjD%N)rR&&QG}p)FwAqXu{43(g)u?^d2u5W zBm>EH;|8>rH=`D8ZAT0uF$trM)h3PI5a~w`?HK<t)Eb@gXWohXK~oYw#P3~LC>}-O zmITg(IR*-=#fhei?jUMIMr*1An!#uX&Pr0`_na!LwKa~%uTsMIlnliqf@4pcf3-9n zmbjE-&L>g@&a;#hRqN{l6}pNidNKa-ywkK_BJCBeCnZaw2cgzP-`%m=+rOiAiB%cA zng=MD=C@XRHNRHWs}v=*+ur~jTUukP{c7KUqN;qgHfyhc5cYUJX}15YY`>xhrDLii z#aY?&hkKC2_$L!pzS^gtsM?nR6`sh4HPHiRTJu--YMozEiZ3Nqf2F7BZOEfN1WH!> z2o&8x#z^-13zV^bFcec|ul6-4Iv^EQ`S$kTEA3ZFJ+)6lQMIpv>?vk;`v-tgOq73f zJ|X87G`_9L-v39CX|?xCJ4JWc0<NYjihj#xzaLY2N{T8yfsRgh6n+dD*76TYdqu5# zyUB;LQ}hX&{W&=~Q*;J(oFyqc1<|{C*7D8!5qV!yq!cB!$L~dHulk>PrJ|(AZShn6 zC$-p5!H|lwSNC&@HY-I*?dAW{X78CT6cwf20M=wLpY9jwvjy^JYu?|g@7~$<m8|Hy zu%-1yWv}k1y?iNYbi6`^CuOJTCqQZJDtmRGtG;7+LDpY&qq0-|_+8kNin3Sty%lu; zkp=}ZkIG)rJQ!Q;S)){~7gUv_94I>_7s8k@Rld59@JM^RK9yxnrLugfZe}U_%S=Qt z%3e`gvsS!Sb(wT%_iNX!&SHOyG@NN`fZcvk7W*$R5w-cX&A=uy`o5eUBg@V!g+1*+ zuqM0iMrm(v=Tcg}M|$4&gHiTYYAZq*t`g)REwUiZ#y_1}R5;Z2%LPpJJy<gxTqD?{ awoFy7Ql!(JO@(L8TPHL=YcsHk?7sjWb7VRI literal 0 HcmV?d00001 diff --git a/tools/testing/selftests/xcall_prefetch/echo_server.c b/tools/testing/selftests/xcall_prefetch/echo_server.c new file mode 100644 index 000000000000..49c1ea035788 --- /dev/null +++ b/tools/testing/selftests/xcall_prefetch/echo_server.c @@ -0,0 +1,110 @@ +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <unistd.h> +#include <sys/socket.h> +#include <netinet/in.h> +#include <sys/epoll.h> +#include <errno.h> +#include <fcntl.h> + +#include "./utils.h" + +static int client_id_counter = 0; +char buffers[NUM_THREADS][BUFFER_SIZE]; + +int main() { + int server_fd, epoll_fd; + struct sockaddr_in address; + struct epoll_event ev, events[MAX_EVENTS]; + struct client_info *clients[MAX_EVENTS] = {0}; + + if ((server_fd = socket(AF_INET, SOCK_STREAM, 0)) == 0) { + handle_error("socket failed\n"); + } + + address.sin_family = AF_INET; + address.sin_addr.s_addr = INADDR_ANY; + address.sin_port = htons(PORT); + + if (bind(server_fd, (struct sockaddr *)&address, sizeof(address)) < 0) { + handle_error("bind failed\n"); + } + + if (listen(server_fd, 100) < 0) { + handle_error("listen failed\n"); + } + + debug_printf("Server listening on port %d...\n", PORT); + + epoll_fd = epoll_create1(0); + if (epoll_fd == -1) { + handle_error("epoll_create1\n"); + } + + ev.events = EPOLLIN; + ev.data.fd = server_fd; + if (epoll_ctl(epoll_fd, EPOLL_CTL_ADD, server_fd, &ev) == -1) { + handle_error("epoll_ctl: server_fd\n"); + } + + while (1) { + int nfds = epoll_wait(epoll_fd, events, MAX_EVENTS, -1); + if (nfds == -1) { + handle_error("epoll_wait failed\n"); + } + + for (int n = 0; n < nfds; ++n) { + if (events[n].data.fd == server_fd) { + int client_fd; + socklen_t addr_len = sizeof(address); + if ((client_fd = accept(server_fd, (struct sockaddr *)&address, &addr_len)) < 0) { + debug_printf("accept failed\n"); + goto fail; + } + + struct client_info *info = malloc(sizeof(struct client_info)); + info->fd = client_fd; + info->id = client_id_counter++; + info->addr = address; + + clients[client_fd] = info; + debug_printf("new connection accepted client:%d\n", info->id); + + int flags = fcntl(client_fd, F_GETFL, 0); + fcntl(client_fd, F_SETFL, flags | O_NONBLOCK); + + ev.events = EPOLLIN | EPOLLET; + ev.data.fd = client_fd; + if (epoll_ctl(epoll_fd, EPOLL_CTL_ADD, client_fd, &ev) == -1) { + debug_printf("epoll_ctl: client_fd\n"); + goto fail; + } + } else { + int client_fd = events[n].data.fd; + // char buffer[BUFFER_SIZE] = {0}; + struct client_info *info = clients[client_fd]; + char *buffer = buffers[info->id]; + memset(buffers[info->id], 0, BUFFER_SIZE); + ssize_t read_bytes = read(client_fd, buffer, BUFFER_SIZE); + + if (read_bytes <= 0) { + if (read_bytes == 0) { + debug_printf("client:%d disconnected\n", info->id); + continue; + } else { + debug_printf("read failed\n"); + } + goto fail; + } else { + debug_printf("Server received client:%d %s\n", info->id, buffer); + write(client_fd, buffer, read_bytes); + } + } + } + } + +fail: + close(server_fd); + return -1; +} \ No newline at end of file diff --git a/tools/testing/selftests/xcall_prefetch/echo_test.c b/tools/testing/selftests/xcall_prefetch/echo_test.c new file mode 100644 index 000000000000..fc7180869f3c --- /dev/null +++ b/tools/testing/selftests/xcall_prefetch/echo_test.c @@ -0,0 +1,52 @@ +#include <stdio.h> +#include <stdlib.h> +#include <unistd.h> +#include <sys/wait.h> + +#include "./utils.h" + +int main() { + pid_t server_pid, client_pid; + int server_status, client_status; + + server_pid = fork(); + if (server_pid == 0) { + execl("./echo_server", "./echo_server", NULL); + printf("execl echo_server failed\n"); + exit(EXIT_FAILURE); + } else if (server_pid < 0) { + printf("fork echo_server failed\n"); + exit(EXIT_FAILURE); + } + + sleep(1); + + client_pid = fork(); + if (client_pid == 0) { + execl("./echo_client", "./echo_client", NULL); + printf("execl client failed\n"); + exit(EXIT_FAILURE); + } else if (client_pid < 0) { + printf("fork echo_client failed\n"); + exit(EXIT_FAILURE); + } + + waitpid(client_pid, &client_status, 0); + + kill(server_pid, SIGTERM); + waitpid(server_pid, &server_status, 0); + + if (WIFEXITED(client_status)) { + int exit_code = WEXITSTATUS(client_status); + if (exit_code == 0) { + printf("echo test %sok%s.\n", ANSI_COLOR_GREEN, ANSI_COLOR_RESET); + exit(EXIT_SUCCESS); + } else { + printf("echo test %sfailed%s.\n", ANSI_COLOR_RED, ANSI_COLOR_RESET); + exit(EXIT_FAILURE); + } + } else { + printf("echo test %sfailed%s.\n", ANSI_COLOR_RED, ANSI_COLOR_RESET); + exit(EXIT_FAILURE); + } +} \ No newline at end of file diff --git a/tools/testing/selftests/xcall_prefetch/mulit_close_test.c b/tools/testing/selftests/xcall_prefetch/mulit_close_test.c new file mode 100644 index 000000000000..fae0407e8473 --- /dev/null +++ b/tools/testing/selftests/xcall_prefetch/mulit_close_test.c @@ -0,0 +1,48 @@ +#include <stdio.h> +#include <stdlib.h> +#include <unistd.h> +#include <sys/epoll.h> +#include <sys/socket.h> + +#include "./utils.h" + +int main() { + int epoll_fd; + struct epoll_event ev, events[1]; + int ret = -1; + + int fds[2]; + socketpair(AF_UNIX, SOCK_STREAM, 0, fds); + + epoll_fd = epoll_create1(0); + ev.events = EPOLLIN | EPOLLHUP; + ev.data.fd = fds[0]; + epoll_ctl(epoll_fd, EPOLL_CTL_ADD, fds[0], &ev); + + if (fork() == 0) { + close(fds[1]); + exit(0); + } + + close(fds[1]); + + int nfds = epoll_wait(epoll_fd, events, 1, 1000); + + if (nfds == 1) { + if (events[0].events & EPOLLHUP) { + ret = 0; + } + } + + close(fds[0]); + close(epoll_fd); + if (!ret) { + debug_printf("epoll_wait detected EPOLLHUP!\n"); + printf("multi close test %sok%s.\n", ANSI_COLOR_GREEN, ANSI_COLOR_RESET); + return 0; + } + + printf("multi close test %sfailed%s.\n", ANSI_COLOR_RED, ANSI_COLOR_RESET); + + return -1; +} \ No newline at end of file diff --git a/tools/testing/selftests/xcall_prefetch/multi_write_test.c b/tools/testing/selftests/xcall_prefetch/multi_write_test.c new file mode 100644 index 000000000000..0af07b742e73 --- /dev/null +++ b/tools/testing/selftests/xcall_prefetch/multi_write_test.c @@ -0,0 +1,77 @@ +#include <stdio.h> +#include <stdlib.h> +#include <unistd.h> +#include <sys/epoll.h> +#include <sys/wait.h> + +#include "./utils.h" + +#define NUM_CHILDREN 3 + +int main() { + int epoll_fd, pipe_fd[2]; + struct epoll_event ev, events[MAX_EVENTS]; + char buffer[BUFFER_SIZE]; + + if (pipe(pipe_fd) == -1) { + handle_error("pipe failed\n"); + } + + epoll_fd = epoll_create1(0); + if (epoll_fd == -1) { + handle_error("epoll_create1 failed\n"); + } + + ev.events = EPOLLIN; + ev.data.fd = pipe_fd[0]; + if (epoll_ctl(epoll_fd, EPOLL_CTL_ADD, pipe_fd[0], &ev) == -1) { + handle_error("epoll_ctl failed\n"); + } + + for (int i = 0; i < NUM_CHILDREN; i++) { + sleep(1); + pid_t pid = fork(); + if (pid == 0) { + close(pipe_fd[0]); + char msg[10]; + sprintf(msg, "%d %d\n", 2*i, 2*i+1); + write(pipe_fd[1], msg, strlen(msg)); + exit(0); + } else if (pid < 0) { + handle_error("fork failed\n"); + } + } + + sleep(1); + + int nfds = epoll_wait(epoll_fd, events, MAX_EVENTS, -1); + if (nfds == -1) { + handle_error("epoll_wait failed\n"); + } + + debug_printf("epoll_wait returned %d events\n", nfds); + + for (int i = 0; i < nfds; i++) { + if (events[i].data.fd == pipe_fd[0]) { + ssize_t n = read(pipe_fd[0], buffer, BUFFER_SIZE); + if (n > 0) { + debug_printf("Received: %.*s", (int)n, buffer); + } + char *expected_msg = "0 1\n2 3\n4 5\n"; + if (strncmp(expected_msg, buffer, strlen(expected_msg)) != 0) { + printf("multi write test %sfailed%s.\n", ANSI_COLOR_RED, ANSI_COLOR_RESET); + handle_error("expected msg: %s, buffer: %s", expected_msg, buffer); + } + } + } + + for (int i = 0; i < NUM_CHILDREN; i++) { + wait(NULL); + } + + close(pipe_fd[0]); + close(pipe_fd[1]); + close(epoll_fd); + printf("multi write test %sok%s.\n", ANSI_COLOR_GREEN, ANSI_COLOR_RESET); + return 0; +} \ No newline at end of file diff --git a/tools/testing/selftests/xcall_prefetch/signal_recovery_test.c b/tools/testing/selftests/xcall_prefetch/signal_recovery_test.c new file mode 100644 index 000000000000..9645f8ffec1b --- /dev/null +++ b/tools/testing/selftests/xcall_prefetch/signal_recovery_test.c @@ -0,0 +1,164 @@ +#include <stdio.h> +#include <stdlib.h> +#include <unistd.h> +#include <signal.h> +#include <sys/epoll.h> +#include <sys/socket.h> +#include <netinet/in.h> +#include <errno.h> +#include <string.h> +#include <pthread.h> + +#include "./utils.h" + + +int epoll_fd; +int server_fd; +volatile sig_atomic_t signal_received = 0; + +void signal_handler() { + signal_received = 1; +} + +void* worker_thread() { + struct epoll_event events[MAX_EVENTS]; + + while (!signal_received) { + int nfds = epoll_wait(epoll_fd, events, MAX_EVENTS, -1); + if (nfds == -1) { + if (errno == EINTR) { + debug_printf("epoll_wait interrupted by signal! re-waiting...\n"); + continue; + } else { + handle_error("epoll_wait failed\n"); + } + } + + for (int i = 0; i < nfds; i++) { + if (events[i].data.fd == server_fd) { + struct sockaddr_in address; + socklen_t addr_len = sizeof(address); + int client_fd = accept(server_fd, (struct sockaddr*)&address, &addr_len); + if (client_fd < 0) { + handle_error("accept failed\n"); + } + + struct epoll_event ev; + ev.events = EPOLLIN | EPOLLET; + ev.data.fd = client_fd; + if (epoll_ctl(epoll_fd, EPOLL_CTL_ADD, client_fd, &ev) < 0) { + close(client_fd); + handle_error("epoll_ctl failed\n"); + } + } else { + char buffer[1024]; + ssize_t n = read(events[i].data.fd, buffer, sizeof(buffer)); + if (n <= 0) { + if (n < 0) { + handle_error("read failed\n"); + } + epoll_ctl(epoll_fd, EPOLL_CTL_DEL, events[i].data.fd, NULL); + close(events[i].data.fd); + debug_printf("closing fd %d\n", events[i].data.fd); + } else { + buffer[n] = '\0'; + debug_printf("receive: %s\n", buffer); + } + } + } + } + + return NULL; +} + +void* client_thread() { + sleep(1); + + int client_fd = socket(AF_INET, SOCK_STREAM, 0); + struct sockaddr_in server_addr = { + .sin_family = AF_INET, + .sin_port = htons(8082), + .sin_addr.s_addr = htonl(INADDR_LOOPBACK), + }; + + if (connect(client_fd, (struct sockaddr*)&server_addr, sizeof(server_addr)) < 0) { + debug_printf("connect failed\n"); + goto fail; + } + + const char *msg = "hello world."; + write(client_fd, msg, strlen(msg)); + close(client_fd); + + sleep(1); + + debug_printf("Sending SIGUSR1 to interrupt epoll_wait...\n"); + kill(getpid(), SIGUSR1); + + sleep(1); + debug_printf("Testing if epoll_wait can still work after signal...\n"); + + client_fd = socket(AF_INET, SOCK_STREAM, 0); + if (connect(client_fd, (struct sockaddr*)&server_addr, sizeof(server_addr)) < 0) { + debug_printf("connect failed\n"); + goto fail; + } + + const char *msg2 = "hello world again."; + if (write(client_fd, msg2, strlen(msg2)) < 0) { + goto fail; + } + + close(client_fd); + return NULL; + +fail: + close(client_fd); + pthread_exit((void *)42); +} + +int main() { + pthread_t worker, client; + + signal(SIGUSR1, signal_handler); + + server_fd = socket(AF_INET, SOCK_STREAM, 0); + struct sockaddr_in addr = { + .sin_family = AF_INET, + .sin_port = htons(8082), + .sin_addr.s_addr = INADDR_ANY, + }; + bind(server_fd, (struct sockaddr*)&addr, sizeof(addr)); + listen(server_fd, SOMAXCONN); + + epoll_fd = epoll_create1(0); + if (epoll_fd < 0) { + handle_error("epoll_create1 failed\n"); + } + + struct epoll_event ev; + ev.events = EPOLLIN | EPOLLET; + ev.data.fd = server_fd; + if (epoll_ctl(epoll_fd, EPOLL_CTL_ADD, server_fd, &ev) < 0) { + handle_error("epoll_ctl failed\n"); + } + + pthread_create(&worker, NULL, worker_thread, NULL); + pthread_create(&client, NULL, client_thread, NULL); + + void *retval; + pthread_join(worker, NULL); + pthread_join(client, &retval); + + close(server_fd); + close(epoll_fd); + + if (retval != NULL) { + printf("signal recovery test %sfailed%s.\n", ANSI_COLOR_RED, ANSI_COLOR_RESET); + return -1; + } + + printf("signal recovery test %sok%s.\n", ANSI_COLOR_GREEN, ANSI_COLOR_RESET); + + return 0; +} \ No newline at end of file diff --git a/tools/testing/selftests/xcall_prefetch/thundering_herd_test.c b/tools/testing/selftests/xcall_prefetch/thundering_herd_test.c new file mode 100644 index 000000000000..bf98fcf94b6a --- /dev/null +++ b/tools/testing/selftests/xcall_prefetch/thundering_herd_test.c @@ -0,0 +1,119 @@ +#include <stdio.h> +#include <stdlib.h> +#include <unistd.h> +#include <sys/epoll.h> +#include <sys/socket.h> +#include <netinet/in.h> +#include <sys/wait.h> +#include <string.h> +#include <sys/ipc.h> +#include <sys/shm.h> +#include <fcntl.h> +#include <errno.h> + +#include "./utils.h" + +#define TIMEOUT_MS 1000 + +int create_client_connection() { + struct sockaddr_in addr = { + .sin_family = AF_INET, + .sin_port = htons(8081), + .sin_addr.s_addr = htonl(INADDR_LOOPBACK), + }; + + int fd = socket(AF_INET, SOCK_STREAM, 0); + connect(fd, (struct sockaddr *)&addr, sizeof(addr)); + return fd; +} + +void set_nonblocking(int fd) { + int flags = fcntl(fd, F_GETFL, 0); + fcntl(fd, F_SETFL, flags | O_NONBLOCK); +} + +int main() { + int listen_fd, epoll_fd; + struct sockaddr_in addr = { + .sin_family = AF_INET, + .sin_port = htons(8081), + .sin_addr.s_addr = INADDR_ANY, + }; + + int shm_id = shmget(IPC_PRIVATE, sizeof(int), IPC_CREAT | 0666); + int tmp_cnt; + int *accept_count = (int *)shmat(shm_id, NULL, 0); + *accept_count = 0; + + listen_fd = socket(AF_INET, SOCK_STREAM, 0); + set_nonblocking(listen_fd); + int reuse = 1; + setsockopt(listen_fd, SOL_SOCKET, SO_REUSEADDR, &reuse, sizeof(reuse)); + bind(listen_fd, (struct sockaddr *)&addr, sizeof(addr)); + listen(listen_fd, SOMAXCONN); + + for (int i = 0; i < NUM_THREADS; i++) { + if (fork() == 0) { + usleep(1000 * i); + epoll_fd = epoll_create1(0); + struct epoll_event ev = { + .events = EPOLLIN | EPOLLET, + .data.fd = listen_fd, + }; + epoll_ctl(epoll_fd, EPOLL_CTL_ADD, listen_fd, &ev); + + debug_printf("Worker %d waiting...\n", getpid()); + + struct epoll_event events[MAX_EVENTS]; + int nfds = epoll_wait(epoll_fd, events, MAX_EVENTS, TIMEOUT_MS); + + if (nfds == -1) { + handle_error("epoll_wait failed\n"); + } + + debug_printf("Worker %d: epoll_wait returned %d events\n", getpid(), nfds); + + int client_fd = accept(listen_fd, NULL, NULL); + if (client_fd >= 0) { + debug_printf("Worker %d accepted a connection!\n", getpid()); + __sync_fetch_and_add(accept_count, 1); + close(client_fd); + } else if (errno != EAGAIN && errno != EWOULDBLOCK) { + debug_printf("accept failed\n"); + } + + close(epoll_fd); + exit(0); + } + } + + sleep(1); + + debug_printf("Parent creating client connection...\n"); + int client_fd = create_client_connection(); + sleep(1); + close(client_fd); + + int waited = 0; + for (int i = 0; i < NUM_THREADS; i++) { + if (wait(NULL) < 0 && errno == ECHILD) { + break; + } + waited++; + } + + tmp_cnt = *accept_count; + debug_printf("Total successful accepts: %d\n", tmp_cnt); + + shmdt(accept_count); + shmctl(shm_id, IPC_RMID, NULL); + close(listen_fd); + + if (tmp_cnt == 1) { + printf("thundering herd test %sok%s.\n", ANSI_COLOR_GREEN, ANSI_COLOR_RESET); + return 0; + } else { + printf("thundering herd test %sfailed%s.\n", ANSI_COLOR_RED, ANSI_COLOR_RESET); + return -1; + } +} \ No newline at end of file diff --git a/tools/testing/selftests/xcall_prefetch/utils.h b/tools/testing/selftests/xcall_prefetch/utils.h new file mode 100644 index 000000000000..58959c564119 --- /dev/null +++ b/tools/testing/selftests/xcall_prefetch/utils.h @@ -0,0 +1,54 @@ +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <netinet/in.h> + +#define ANSI_COLOR_GREEN "\x1b[32m" +#define ANSI_COLOR_RED "\x1b[31m" +#define ANSI_COLOR_RESET "\x1b[0m" + +#define PORT 8080 +#define BUFFER_SIZE 4097 +#define MAX_EVENTS 100 +#define MAX_MSGS 4 +#define NUM_THREADS 4 + +#ifdef DEBUG +#define debug_printf(fmt, ...) (printf(fmt, ##__VA_ARGS__)) +#else +#define debug_printf(fmt, ...) (void)fmt +#endif + +struct client_info { + int fd; + int id; + struct sockaddr_in addr; +}; + +void handle_error(const char *format, ...) +{ + printf("%s", format); + exit(EXIT_FAILURE); +} + +char* generate_number_string(int length) +{ + if (length <= 0) { + printf("\nnumber string length invalid\n"); + return NULL; + } + + char* result = (char*)malloc((length + 1) * sizeof(char)); + if (result == NULL) { + printf("\nnumber string malloc failed\n"); + return NULL; + } + + for (int i = 0; i < length; i++) { + result[i] = '0' + (i % 10); + } + + result[length] = '\0'; + + return result; +} \ No newline at end of file diff --git a/tools/testing/selftests/xcall_prefetch/xcall_prefetch_test.sh b/tools/testing/selftests/xcall_prefetch/xcall_prefetch_test.sh new file mode 100755 index 000000000000..e9a990dac69a --- /dev/null +++ b/tools/testing/selftests/xcall_prefetch/xcall_prefetch_test.sh @@ -0,0 +1,39 @@ +#!/bin/bash + +RED="\033[31m" +GREEN="\033[32m" +RESET="\033[0m" + +make clean +make + +if [ -e "/proc/$$/xcall" ]; then + echo 22 > "/proc/$$/xcall" + echo @22 > "/proc/$$/xcall" + echo "enable xcall epoll prefetch, pid is $$." +else + echo "Warnning: not enable xcall." +fi + +./echo_test +sleep 1 + +./multi_write_test +sleep 1 + +./thundering_herd_test +sleep 1 + +./mulit_close_test +sleep 1 + +./signal_recovery_test +sleep 1 + +if [ $? -eq 0 ]; then + echo -e "${GREEN}tests passed successfully.${RESET}" + exit 0 +else + echo -e "${RED}tests failed.${RESET}" + exit 1 +fi \ No newline at end of file -- 2.34.1

2 1

[PATCH OLK-5.10] add xcall epoll prefetch selftest
by Yuntao Liu 09 Jun '25

09 Jun '25

--- .../testing/selftests/xcall_prefetch/Makefile | 37 ++++ .../selftests/xcall_prefetch/echo_client.c | 119 +++++++++++++ .../xcall_prefetch/echo_client_multi | Bin 0 -> 21896 bytes .../selftests/xcall_prefetch/echo_server.c | 110 ++++++++++++ .../selftests/xcall_prefetch/echo_test.c | 52 ++++++ .../xcall_prefetch/mulit_close_test.c | 48 +++++ .../xcall_prefetch/multi_write_test.c | 77 ++++++++ .../xcall_prefetch/signal_recovery_test.c | 164 ++++++++++++++++++ .../xcall_prefetch/thundering_herd_test | Bin 0 -> 21000 bytes .../xcall_prefetch/thundering_herd_test.c | 119 +++++++++++++ .../testing/selftests/xcall_prefetch/utils.h | 54 ++++++ .../xcall_prefetch/xcall_prefetch_test.sh | 39 +++++ 12 files changed, 819 insertions(+) create mode 100644 tools/testing/selftests/xcall_prefetch/Makefile create mode 100644 tools/testing/selftests/xcall_prefetch/echo_client.c create mode 100755 tools/testing/selftests/xcall_prefetch/echo_client_multi create mode 100644 tools/testing/selftests/xcall_prefetch/echo_server.c create mode 100644 tools/testing/selftests/xcall_prefetch/echo_test.c create mode 100644 tools/testing/selftests/xcall_prefetch/mulit_close_test.c create mode 100644 tools/testing/selftests/xcall_prefetch/multi_write_test.c create mode 100644 tools/testing/selftests/xcall_prefetch/signal_recovery_test.c create mode 100755 tools/testing/selftests/xcall_prefetch/thundering_herd_test create mode 100644 tools/testing/selftests/xcall_prefetch/thundering_herd_test.c create mode 100644 tools/testing/selftests/xcall_prefetch/utils.h create mode 100755 tools/testing/selftests/xcall_prefetch/xcall_prefetch_test.sh diff --git a/tools/testing/selftests/xcall_prefetch/Makefile b/tools/testing/selftests/xcall_prefetch/Makefile new file mode 100644 index 000000000000..5bd509df66e0 --- /dev/null +++ b/tools/testing/selftests/xcall_prefetch/Makefile @@ -0,0 +1,37 @@ +CFLAGS += -Wall -Wextra -g +TEST_GEN_FILES := echo_server echo_client echo_test multi_write_test thundering_herd_test mulit_close_test signal_recovery_test + +# enable debug +# echo_client: CFLAGS += -DDEBUG=1 +# echo_server: CFLAGS += -DDEBUG=1 +# multi_write_test: CFLAGS += -DDEBUG=1 +# thundering_herd_test: CFLAGS += -DDEBUG=1 +# mulit_close_test: CFLAGS += -DDEBUG=1 +# signal_recovery_test: CFLAGS += -DDEBUG=1 + +all: $(TEST_GEN_FILES) + +echo_server: echo_server.c + $(CC) $(CFLAGS) $< -o $@ + +echo_client: echo_client.c + $(CC) $(CFLAGS) $< -o $@ + +echo_test: echo_test.c + $(CC) $(CFLAGS) $< -o $@ + +multi_write_test: multi_write_test.c + $(CC) $(CFLAGS) $< -o $@ + +thundering_herd_test: thundering_herd_test.c + $(CC) $(CFLAGS) $< -o $@ + +mulit_close_test: mulit_close_test.c + $(CC) $(CFLAGS) $< -o $@ + +signal_recovery_test: signal_recovery_test.c + $(CC) $(CFLAGS) $< -o $@ + +.PHONY: all + +include ../lib.mk diff --git a/tools/testing/selftests/xcall_prefetch/echo_client.c b/tools/testing/selftests/xcall_prefetch/echo_client.c new file mode 100644 index 000000000000..2e3c41b2b14e --- /dev/null +++ b/tools/testing/selftests/xcall_prefetch/echo_client.c @@ -0,0 +1,119 @@ +#define _GNU_SOURCE +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <sys/socket.h> +#include <netinet/in.h> +#include <unistd.h> +#include <sys/epoll.h> +#include <arpa/inet.h> +#include <pthread.h> +#include <errno.h> + +#include "./utils.h" + +#define DEBUG + +void *client(void *arg) +{ + int thread_id = *(int *)arg; + int sock = 0; + struct sockaddr_in serv_addr; + struct epoll_event ev, events[MAX_EVENTS]; + const char* test_messages[MAX_MSGS]; + + if ((sock = socket(AF_INET, SOCK_STREAM, 0)) < 0) { + handle_error("client:%d create socket failed\n", thread_id); + } + serv_addr.sin_family = AF_INET; + serv_addr.sin_port = htons(PORT); + + if (inet_pton(AF_INET, "127.0.0.1", &serv_addr.sin_addr) <= 0) { + close(sock); + handle_error("client:%d invalid address\n", thread_id); + } + if (connect(sock, (struct sockaddr *)&serv_addr, sizeof(serv_addr))) { + close(sock); + handle_error("client:%d connection failed\n", thread_id); + } + + debug_printf("Thread %d connected\n", thread_id); + + int epoll_fd = epoll_create1(0); + if (epoll_fd == -1) { + handle_error("client:%d epoll_create1 failed\n", thread_id); + } + + ev.events = EPOLLIN; + ev.data.fd = sock; + if (epoll_ctl(epoll_fd, EPOLL_CTL_ADD, sock, &ev) == -1) { + handle_error("client:%d epoll_ctl add sock failed\n", thread_id); + } + + test_messages[0] = "hello world."; + test_messages[1] = generate_number_string(100); + test_messages[2] = generate_number_string(1024); + test_messages[3] = generate_number_string(4096); + + + for (int i = 0; i < MAX_MSGS; i++) { + const char* msg = test_messages[i]; + ssize_t msg_len = strlen(msg); + if (send(sock, msg, msg_len, 0) != msg_len) { + handle_error("client:%d send failed\n", thread_id); + } + + debug_printf("Client:%d send %s\n", thread_id, msg); + + int nfds = epoll_wait(epoll_fd, events, MAX_EVENTS, -1); + if (nfds == -1) { + handle_error("client:%d epoll_wait failed\n", thread_id); + } + + for (int n = 0; n < nfds; ++n) { + if (events[n].data.fd == sock) { + char buffer[BUFFER_SIZE] = {0}; + ssize_t read_bytes = read(sock, buffer, BUFFER_SIZE); + + if (read_bytes <= 0) { + close(sock); + if (read_bytes == 0) { + handle_error("client:%d server disconnected\n", thread_id); + } else { + handle_error("client:%d read failed\n", thread_id); + } + } else { + debug_printf("Client:%d received %s\n", thread_id, buffer); + if (strncmp(msg, buffer, msg_len) != 0) { + handle_error("client:%d error: response does not match sent message\n", thread_id); + } + } + } + } + } + + close(sock); + pthread_exit(NULL); +} + +int main() +{ + pthread_t threads[NUM_THREADS]; + int ret; + + for (int i = 0; i < NUM_THREADS; i++) { + ret = pthread_create(&threads[i], NULL, client, &i); + if (ret != 0) { + perror("pthread_create failed"); + continue; + } + } + + for (int i = 0; i < NUM_THREADS; i++) { + if (threads[i] != 0) { + pthread_join(threads[i], NULL); + } + } + + return 0; +} \ No newline at end of file diff --git a/tools/testing/selftests/xcall_prefetch/echo_client_multi b/tools/testing/selftests/xcall_prefetch/echo_client_multi new file mode 100755 index 0000000000000000000000000000000000000000..36aa08d6e12f6e8ec73b2a7f8eb567225eb0b2f8 GIT binary patch literal 21896 zcmeHPYjhmNm98GCr5TMLBTKd=Ti9cR9ShLQFKmpBY-{is%MWY|>@4hfq-jY5qZwr$ zevlX&WF?*mWo;G$$@&mO&LIiO$uT*B@G{7}Y-bbJ>=JMun>7jKNWo-{fJ8uG_Ph0H zwK|&JJvsYpqta2|d%wE(R@JSV?&_}U-sfw+#-V9Ur5v_`k#ug7KpJJn!Dh*TG_obE z2**p<#VikUF2_WEqaer?>8xWcbxJ%3l<X>q8HZL0m?^4iNR;fdC68O=D9SuSCOe&2 zDcfmLfzVS_<(uVsMLx69a#?>xt#%ZfiW*^8v3x8Yk#S+tUfB^$cJrj&JZY!s1*#B| zqAH#g6Z&5w{n=?f86hcZx9gO4cG@J&n4(k$r7dqILw`+qo21>qEaAsa)i6_3m3JNN zC@%loq`G;tEH7I>49b2`REC$;MZ&F%7uH2WHIZ;MwYz5blEpQP7uF_Xwexuus9s+D z)0kSbp@qTC0AZ>-9s7l>{83)H%qRWs(jWZ%*E_C1J?FB63+}pchWni>DsCnnsyEq? z4khxZh6>5VPs2ayh<Mx%PGCe|F<$N>nk}tBhWu%dQHf{4D^QSkjHVE7&7yyI7W}0w z_|Yu*KV`vh$%608f(NtU7iGa0W@+yYS@fyD#*63dEcioN@ZK!=BU$i@EcoIq`06a} zs?MVScouw9798#wwfQenAkelu7-$bigOTv92J23C#*JVoVC)Vj8F8#D7>UH%%#533 z;V5g1MWaSrk_7_z8MXxy$zVJg=n5hy5o^1}NV0I$NCvu-u_!aTW06RpEg4~5Mppt^ ziDW$5*453rQ^^Eti^LM6g|?x<q(K@H1Fno{h;_n8LOR(MM4h^gcsv$YZn;4N>VYb> z-4bZ)yd}^c3`ZDMi%tfM*_!6Ht5yf**Df`=1vo`0bfL-3uU%yF3l_4#+O6vYAtP>d zgcC_4zIFZTNGxh>4Yo!Ml-AJ|i^^69gzC782(|-f2o3)^IQ!F){&f5cjCj(ZKMh<S z=ctb~<h%~5&Xbkla1otR-jG<lGVG$W+uKqu=lQwpPf|`zqkMQp$CHmA9{ws$M-Em* z0|7~0@6>f>piJOic|BA3eu+C9MZUt7eD-RA)45K%=D;}%uFkoXF=)ZjATwRC;8RSL zu@MVS=T(*JC7ut!dj4Hv!RdUhQlkau^9-5cx8T$Ul{Q&$Dnq4h7F@-G@^)Bo1l~-Y z7F<qYcwV;!ms28+@3!EDrWeNcTkv8Fe$axKSnxv@T#X^hzR!Y}TJ(=vaO?T<+ZJ37 z1#WfBg41<VrN=G!WCel_Sn#P9{FDW!bEZmXEO?~?C+I{3CL%Bqfxn>$yjL>gBY)3% zr@v4Cc^zZ^-hreebH?BEg7bMEh0LOlY8e~O%=s9kWF{m1PRbb`dLxs`9O5!vn1_d6 zF=e`74iB9$Wx7BQ4?SYabm1Hxdcc(Ff;l{N*p%r)IXrZ`DbodVc<6JcOc%!Cp`a<# z1#x)jI#Z?#;qcH(Q>F{x@KA#((}iz%XqG9{L}7TS)RgH0I6S18GF|wFhu-~EwU;h< zXum1bg%0gEWxBwj{iaM8HniWA>4Jv#D;d%4;?ZqlJid_csiA0(%KbLE*Cy|^$tj!M zWs?n?e3MOXvB~Rgvd<<jx5<laa*a)%W0Nnk$&+nzkxkZZ@+X&%jn@ZOx#69)=~w>V zpZ<-%=gkY7w)z?d8czHBmtTQlmYMu-(0sqWwq)i3bP9d|{mVZD;_ugoi7XvSPR38; zG2-#laAnSw%-m0U&r1sZ(_DYib402?^`|fRPyFF(|A~<tzxI;<l~0r9aL_9qI5X$k zdA}-q`u*%*PT4GVS&P4Cc`LF0^jpb7f8X+U2!lWWB$F8oVKBa=-vn8^3ASeWWIwbE znZ&ff&VT#)0q)_rAE-b5=r&I7*}+MFCrER;x0{px-5>`BlHQVP-~N*7^@mETTaK1g z-*l{`+88LQ?mAOaojO-iz4yYi^a~%{fdOk4({tXN-g~}h?|HUw3WD%7eq)0hQB%5c zWJCAj(C0hfGobnJ^QTV^zJ(lr+Bf3w**n5gWia>-8O%@yZ--C%-Y3hqoBCSb_qRRg z4-a&-x3~LG40=zDIMT13_;hCVh4hK^sT1cM+9_=y{Yo{2et&(R59(+9eekWFN&onu zr^olc25_10{1<+RDvV~ngt$WO^z&vLZa+`=F8-dvOBlk|h|`cief-q#GntUTFL%y7 zREO%^FyQa&CA5#Fd(VQN?EMvvR9pYa-V6U}vLlM^zlU9rk$B=$gwCIS{XPG8{ppVm zI{ZD)X-|^rGZfNi=}bB3M-<ZBXU}o&c~4~@FDU&I0`(xyW<02am56`)$zIRfnT&%y zduS$1i~bBK-8%`8zn|y$dj@j->F50Yi_$O@1+0Nq@4&vfeY_BKM;-u65n;24$EcsA zqx1=(|0JF6{OOlr7v+wUqmGV$JjT&7(@_^3JqtTF_*vMbd(R<aJ;&*oNm*NWxoNQs z7AJdi(GrS`tZJ{5x2&JwzI!M*ecTm~9xAjR<@bz?b@?GK2O?mDC>p<HdZz17_j=*u z*$`Ff3TkD#cOC)s)O2vJ0O~n_<_=C4z!Cr_d(X(;9(n^c>*FFub>D#($s_y>A|FG9 z6Vz$@M;yJt(R&1Wlrlw|o}j2et(DZ4_ELV*8j*-u{c2wQM;lR!X!R3_Ugk+^0_-1z zeY*EenD-o~<Jcf9H@*B8Av)RnDm19zp<Lbu66u$bKI7L0@Wb)rXXV!h{WVYH+SJFE z@_|y)dj=L54m8G})DV|v$OZkLAN>&a{r!|72MJ}1L2{b%(!C>8M9*;|Y~Om>pC<?V zC|AO%>6StGFs|yN+Zs^gx3^iczHeqtp{(WpJ`q+?0W<V3Q_Y?lp|L(Dq6ba$*AcTG z9t*9HlVv=BC3pTFKdm0#vs%w_>hGMAuS^1FR{K@B72~-84KU9jJ?Fb^=ZcfQyD3O- z<5Y9EzrW=uM#_nEIsX2u9R7YF=@b6`$)86ncqj+of!nn2NYCCQELnsfdpW8)I0*0% z%yITi`|c)Y#C6-TlfL`7<4gN|_tEfcdhsY-lBSM|X79WX2R**~w4S~9xo$hdgX-^F zKN9j^R>;q4Nqz8zKV>qfF~y^n@cExP<D)_*(nJI%A}|qwi3t3Ej{vO-P%4O~x>}96 z7i%NoXonZ;BOS?3Z#cRW%Qhi<u2{11(n^Ryvk|1WNZ5!bubdO|ibWQ$Sg4UY1*~EI z6}9#FYp`fa@7`c26gLuy0!x-$01C&VrU!dDa^Xj;(3o1I^)MHP$Um=+snls;$;rDb z7LSB#Errs8kf~#LM2kaarnS8ZBfb-X3xyN1(niSI5?bRL*N(3<UFk*3yRigictbHG z;f=<U-mYM>t<#IjfJIn>9Y%q<IwpfV8hLGc&a!5N2J;5C;}4n4e$WBT%DxSH26O<_ z`)(%F3F;coWX^(igVN2`!uK+na!dzzVlGfmlWrtT|9H}F-ORMz9&LJIzVi_B2&cUm z$8cjamo%F4G2vs_0|GzzKl@H5vx9sTd#)+2xVFT-%ekLjJ?+X%7hEzMD7U{6)Pp*4 zLjsY=Zxa7b;G1L}2(Q5Zo%pXt-8%^I6??wuSY1SqhNvFYz9jNq05$U*PocKL_7va> zv~yLiB_Ax=zKQ=8f6Qb^XVmstZcETanux$e1STRd5rK&aOhjNJ0uvGV+lzqOucY=T zsi`ULnW9uEh`E8PNXc>&k=j$D_CZaT`Iw&ar1l!A{ZKS7rKIvd_%stE`Iy|?r1li; zpb0RNB<!rr$0Kf@Zju#MdwyshP06}fhGc2RDy}AJbdS@Bw5NGFCABw6?Ww7twN?wh zO;%Xddw}QSUPDq<{tPKAJv;&B?L%=K!n0N`kH`Y?JX_^+OWbyF`5Bpyhp=2$?U3&O z&xqP@ZFg8NZ(bTD-6ZJ_NxLQ8FX<skk4kz>(g8`&NP14v3z9OqdCDWHSJHY(8ztQ& z=?+P|CAHW4f0CVzLSxP9)mM7wwX~+9$&|ODzIIV<eN97(iw(EUpI=+Quv)_Ga#)Cm z2clC?%l%u6)Y!C@IUQ<rH-;Y92GH+?m*P+dWIRKMGye#jJ96_M0LgJWG-n&hj{LWY zXn6)m?!EAkGj}TODRemW7eR7qaV=+FRTGGwtB0dWXc2Vj6uM1yIQ2&`WO6s+kW(`q ztH><3jhs9;{Q&qp=XXd+%ZtE!KJj_ZW29q1CwF}rd4h}Y!5S63m#T084bZ1^PtE}F z{9G!hpdzmpCa!77)m;0*I$T{;rt5QL<SGEKyB<Prp6kot3koW6q~C)3XHWpmrG=Wj zD&CO7xlr`v|IpbAtmsu}=0B#Be^&Hcs_2gjI;aGee<esUReKkh5^7x$T0p3WkkVh1 z=^fzmoOJZuN#HR!&!<S{IqB$mfWS}5@Ao)JN6%vft|Idj9HgV?pK&PsI@QhjfwKY@ zMHi^<f0A|kCe)oHCGUbbipZ<;!}8Z4<Q9Dey3RkB?S!e5<-f)~($RAfGA4Z=hbm{T z=G+UW>;}@!(}d4554p+LgwwJIsW_)r`~&h@Mm^<pX_Nns+?M@@qymvw!Sd(9O`emE zo;z?Ty%yQ|{T!sDhdQ(LT6oWYkb`vee2+jHUHL!aARRp?2wXv$Kj9!9J+BeiN8lSA zq@#z1YUx)|xBNeHkd7WY_2%)O^td_uC0Kc;aP|nXvpD+*v6pl9C1S7OECty^-H^wF z<5|zyy9nLJ*>4fs#@VNdjdJ!CV)t<Ncf=mx?8n6Z1K8ZJp#)b&*=Dpf_h}qlmE{KU zUn726`EKH`gQ9DCc^~l&;GHujQP>L~gl(>Krjxq4=p<Y^XO;XIJWwLe&rf;?q}=2D zcl|PS%;Z~1<v$8Ie*<it=iJK(|2-+c;a&%x6@MK9p?nONuwrsSIb6<Xyu+!Rxtzla zAAym_`Aa8_sG@1)^*63hU;|gweP2|n`OMh`Ui?Vi7w6^Ry6XBDXlSlp)XU-8fMc$! z3mzI>x52@FH@`A#Zhp0KxVMwfT-OO?Ib9!v_qh3#z+3PzY)CHn3JO5-7PNB88Vh#v zny~$B%3QK@H<!>}gQ?xgICF1sErZ(B_7HP#%>Nv*sRi?yds86=cWP?!Lgv1%@ELND zTY`goa}L$H@F~=@(7n}}gi!P(vfbBn3syuO?7qQ6CFHV-yI|umbaK2{&Ph&zr=%VE z7&?`+QEx{_3Gq3H5AB;4#~yk|r;99d4&S+N`gPKbw;bRvpg1ke95)vWl>_@&)x%O} zDfvaB{J6|NxDTE1Gk|Vf7kGcU*hPD&(9(Qr`ulRK!JpJY){X>mjh_*WcR-_@G;$6< z-cq>$jH66e>r_kSGl;UIT*^Obsl1is$rW@K$vM3L`pRX9iDOCu`N}zbE-=m5$sAKF znt3bVZ>gk^Ii^;{dG@jED|5)aqVx_fAG*G>l;q2#`!g+7v*iTn6Vc1Og;$k`2n@7T zHOTB=sO+u;s=s{lqrCcOTX=jND@vc@^0}7E&r<omBH`fGma5-L2XmBzD05spNo2o( z;^jQ*4D2aY&cng6p;Gu8Y~dh3^HYmv(wQG;nqq1$HHnA4jTKXM?ox6NyXf<fwb}XK z1gV$Z@B(;B+L2)>dY?djwAr(%B96m1R??MTt9Fxt<8WtHS{jM+dB1CQyx$#%LyXP( zBU-6d<*5sN2h1pGN4zg%lxkHoMSLE=Z8oG*ZBB8SJEwShaj7;b--CcSE@5u>3P9+? z60jK2Q&>*%mF{KkrS7$uZs<}0)yg9U!t&=bBDj9!6t8d-CiFFcu6Z!8f}aA;U_iP# zgD@4ALEr5j)h#Nr6tx64#hSLZjJ%Z0LJ4(1JXi3dO3R>m6}otm=|f-YCe1SG1LE2; z@=-ojvz9U0S_WfNrrPQ-75UE5=2p~dS67a<pTg8w@zqt8R8-|cOLZL>F4F**wggOf zBgB*PNx4d}j9p~)_?gPJWn+BKz?&pvT+HMxN3^`iG5dj>&l>B<bc45o_|W|+`a6eB zq0yv$TwH{sMkAf%T2_}z#OuP*wn!>u)ZtCLz~Y7SwY!dJs;)JR7kR8IV}v@q2}ExL zjRpV0Nn%d|JM5C-NTRlrN;TR#V*#<o6X;4slHuAm8kxyG-3Do(wuxjY90Q9XEMHo( zX5uv@TDFp^{G}wphohi4d60QDS(Czn5OT=10^~&ie1!@u;>mQ4xg6?ul2fJWy5pvD zU3*Y>zNNc5bk9@zr4Q)S+jZybx+_Iu@h)9Y>N7Ay&Z*NYi8wBuq?Zu0O~|M8GLmZb z*+gn|yb6@7U8}plsF$D83s>mcx?D%jjkS*1lXTZxZ|Ma`^-{;;DZ0~nNOv99r#TNC z)+_GUT|0rirn}aib{<fiUhe#o?tG8bN}RpA=Nr1~DUr8bp8}xbb$!bHdPQpVSh7pc zyI-HYU7zmk&`aOci&A=_6XY;=Rp_{+jQB~gb=|Me+^&0^!+Pn{`dnv^UhcTKT%Y_6 zz2dZ9`3>h$=Yx6$EGyUP7o$CnoNnhGx{F$t)aP&@?)2(;0?>2joH~slKB*TEQncpQ z=_Oa`g^p^J*72178Pc!Pr;_p^5uU5`;vJ+}UZ)q{uNQ3>I?j*^USUMnwmYU^_~HZ~ zpts1deuv>B-xmu7lfgie`8I8AZoXmjhK-xo^LN|=t$UJ2g2`9ku%1dofrd{YoQPS3 zXfWO_jMr~jBbbdFd|Uh*w<?#bR&How9Y)lM<3+ZBShK~8Vzg$97wPbRTp&OrZoz!i zA`y-*4zx#NyTZ|SltL1fDODO4lS*eSp2RAztn%6oqMXf5{+3O`0*jj6Mu4guj^fp~ zu5e_J(BT?*v5!Nlv~IYAe<~^^n~Rt0#`&q40l>?5CLiwJO_dYXY*-9Fk-$5HDxjO2 zctFLEmTlR%dR?Gt&E}QsM;)Y8`<I9ciT3i|@NHY&+_Giu^*$z+w&?|1@9Kst<jt(N zrYo6>8q0ZT!)*|Q@wU$8;;mnAO^3IpJ($=NZR?E3qOnw>CKW|gLp4cyUoU}7{(4|d zcRZHFYm2dX4Z0+TWptI%76~RgRW>bZ*R;2_TV&bSEn8)$i8!aCiEu~Mz><1rFfJOr zxoQ1qh?WSmO`F&9-nrF?#~9us#OggAOl#`i(HKV;%du+g#0oP@h#0lU;#hVTWo+2E zVO2At%M!r=4QM$^#`hPB2%{~&>o%=iz0S8)h^w17`nLI2TYeUJWs?_$o@q_Bw;S=% zIvIG4Q;gfi3y~dd4<)GYM^Bfs?JI8(4MrinPd7L5p93EWvT(DVyXP<BLE$5EG$<4c z-tdHUwRtmDjI?OTsPQ~nHw48Tl9-ow2BV>f5fJZ);x~=AQpFOzdWRJAcwV0^TQ~bG zLzC5Y#=4BUNPYd{h4YszTv8W{+_I>lzAhPyMG|%72*2REgb`^cK2f*34R6H;y5mN> zfz1G-t^wcmz71Qq;JwxOPSI=XWm<K{S-H+fzgzR<5*2l6TD{tW26<?pL)W{ki7w8$ z>|MTR)T%_!^)%$%iOV7-FF#_Y13I2QJ@<29D(WExOY(vzywQd)khps8s!Y~`(eUQG zr%@nOKT0(7)YufJo}+RQhcDot{14dtALROz&G%E4{<px4SAQ}e52uN@hTRLEi<O5_ zI-dS%u0NI8pDX?=H!EWUvqtTSdkg;>c0s7;vnuyvFq4#Eq4b{qc=+YOy%s&i>X?vt z8Ro2w3K7(w#m{!&9_Xw1E2|x(?<mD|Qz3%3XYoTfe&f~i?kxC2S@0ia!GD|u|5X-z zI18SG_)vT8@$vAujcqGQGl7>nDw%qYsx&SqoLcM^2xf*nEdxGY9InfP-<$=fM`RuZ zzoJn(mc+&NJ?3{@=1QENeUZMsAL)tQB;?p168s~_$NDDe;Vk$I(vSUi@|Rik{~-14 zPXV@P(f=R|?!x(q;%{%)WZ>SvtUjv)K3@FSWWnjt?(A{=+)DZmtOj|;X71;HJm#}) zb?e)o#XsB97Ed<R#&ER>wuS@AU<U(`Na9k7JCo551q6~^xNV6=F`WzqLa{(cB-R>? z1VWf};8LF2jZY_Zbw`Y(5vr{pUxU8I5Do<6@!*~SW&rU$tUVs=G6JDgSJxhxSR_p3 z!pSkJ<IUv)fom`e^96hxngaM*LDTjPE7z}G4NsV21$=&)>u=i30&AK#u3Fg~*m%t~ zTYOssTUV}X_QAsbk%vb0^#pt;0TU_unu7_^XB7UsZzs@KAI2&hd=A4~P2Mp58=G(Y zw1b(sP<$K0l&NIeAvgvQ-<%i&p?4C-7{uI(XrlCq2-+BE0=!v#wk1CMfS`>121Fo~ zhy}##1MS}kE!<uPQVHz8G{1#0#+#ZO(1$k0WQfh1V<3!!5K3%KBxLFQRzg&p_C1=) zRBz>2Y(*T+TA1y=M0`=hRHUZLIlSC3IVLL*z_&2eSfY<)jQiyaRSEi9h-r@DB10`U zg_;>l(Kzc@NX!f#b{gbkCw5}$IYx`sCiZk;x(}L+3)-nTOeziBxYc6wrcsM07PYt! zK%<j;EKqAIjBg3yK1yh<T(uTA4;_rFcjD%N)rR&&QG}p)FwAqXu{43(g)u?^d2u5W zBm>EH;|8>rH=`D8ZAT0uF$trM)h3PI5a~w`?HK<t)Eb@gXWohXK~oYw#P3~LC>}-O zmITg(IR*-=#fhei?jUMIMr*1An!#uX&Pr0`_na!LwKa~%uTsMIlnliqf@4pcf3-9n zmbjE-&L>g@&a;#hRqN{l6}pNidNKa-ywkK_BJCBeCnZaw2cgzP-`%m=+rOiAiB%cA zng=MD=C@XRHNRHWs}v=*+ur~jTUukP{c7KUqN;qgHfyhc5cYUJX}15YY`>xhrDLii z#aY?&hkKC2_$L!pzS^gtsM?nR6`sh4HPHiRTJu--YMozEiZ3Nqf2F7BZOEfN1WH!> z2o&8x#z^-13zV^bFcec|ul6-4Iv^EQ`S$kTEA3ZFJ+)6lQMIpv>?vk;`v-tgOq73f zJ|X87G`_9L-v39CX|?xCJ4JWc0<NYjihj#xzaLY2N{T8yfsRgh6n+dD*76TYdqu5# zyUB;LQ}hX&{W&=~Q*;J(oFyqc1<|{C*7D8!5qV!yq!cB!$L~dHulk>PrJ|(AZShn6 zC$-p5!H|lwSNC&@HY-I*?dAW{X78CT6cwf20M=wLpY9jwvjy^JYu?|g@7~$<m8|Hy zu%-1yWv}k1y?iNYbi6`^CuOJTCqQZJDtmRGtG;7+LDpY&qq0-|_+8kNin3Sty%lu; zkp=}ZkIG)rJQ!Q;S)){~7gUv_94I>_7s8k@Rld59@JM^RK9yxnrLugfZe}U_%S=Qt z%3e`gvsS!Sb(wT%_iNX!&SHOyG@NN`fZcvk7W*$R5w-cX&A=uy`o5eUBg@V!g+1*+ zuqM0iMrm(v=Tcg}M|$4&gHiTYYAZq*t`g)REwUiZ#y_1}R5;Z2%LPpJJy<gxTqD?{ awoFy7Ql!(JO@(L8TPHL=YcsHk?7sjWb7VRI literal 0 HcmV?d00001 diff --git a/tools/testing/selftests/xcall_prefetch/echo_server.c b/tools/testing/selftests/xcall_prefetch/echo_server.c new file mode 100644 index 000000000000..49c1ea035788 --- /dev/null +++ b/tools/testing/selftests/xcall_prefetch/echo_server.c @@ -0,0 +1,110 @@ +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <unistd.h> +#include <sys/socket.h> +#include <netinet/in.h> +#include <sys/epoll.h> +#include <errno.h> +#include <fcntl.h> + +#include "./utils.h" + +static int client_id_counter = 0; +char buffers[NUM_THREADS][BUFFER_SIZE]; + +int main() { + int server_fd, epoll_fd; + struct sockaddr_in address; + struct epoll_event ev, events[MAX_EVENTS]; + struct client_info *clients[MAX_EVENTS] = {0}; + + if ((server_fd = socket(AF_INET, SOCK_STREAM, 0)) == 0) { + handle_error("socket failed\n"); + } + + address.sin_family = AF_INET; + address.sin_addr.s_addr = INADDR_ANY; + address.sin_port = htons(PORT); + + if (bind(server_fd, (struct sockaddr *)&address, sizeof(address)) < 0) { + handle_error("bind failed\n"); + } + + if (listen(server_fd, 100) < 0) { + handle_error("listen failed\n"); + } + + debug_printf("Server listening on port %d...\n", PORT); + + epoll_fd = epoll_create1(0); + if (epoll_fd == -1) { + handle_error("epoll_create1\n"); + } + + ev.events = EPOLLIN; + ev.data.fd = server_fd; + if (epoll_ctl(epoll_fd, EPOLL_CTL_ADD, server_fd, &ev) == -1) { + handle_error("epoll_ctl: server_fd\n"); + } + + while (1) { + int nfds = epoll_wait(epoll_fd, events, MAX_EVENTS, -1); + if (nfds == -1) { + handle_error("epoll_wait failed\n"); + } + + for (int n = 0; n < nfds; ++n) { + if (events[n].data.fd == server_fd) { + int client_fd; + socklen_t addr_len = sizeof(address); + if ((client_fd = accept(server_fd, (struct sockaddr *)&address, &addr_len)) < 0) { + debug_printf("accept failed\n"); + goto fail; + } + + struct client_info *info = malloc(sizeof(struct client_info)); + info->fd = client_fd; + info->id = client_id_counter++; + info->addr = address; + + clients[client_fd] = info; + debug_printf("new connection accepted client:%d\n", info->id); + + int flags = fcntl(client_fd, F_GETFL, 0); + fcntl(client_fd, F_SETFL, flags | O_NONBLOCK); + + ev.events = EPOLLIN | EPOLLET; + ev.data.fd = client_fd; + if (epoll_ctl(epoll_fd, EPOLL_CTL_ADD, client_fd, &ev) == -1) { + debug_printf("epoll_ctl: client_fd\n"); + goto fail; + } + } else { + int client_fd = events[n].data.fd; + // char buffer[BUFFER_SIZE] = {0}; + struct client_info *info = clients[client_fd]; + char *buffer = buffers[info->id]; + memset(buffers[info->id], 0, BUFFER_SIZE); + ssize_t read_bytes = read(client_fd, buffer, BUFFER_SIZE); + + if (read_bytes <= 0) { + if (read_bytes == 0) { + debug_printf("client:%d disconnected\n", info->id); + continue; + } else { + debug_printf("read failed\n"); + } + goto fail; + } else { + debug_printf("Server received client:%d %s\n", info->id, buffer); + write(client_fd, buffer, read_bytes); + } + } + } + } + +fail: + close(server_fd); + return -1; +} \ No newline at end of file diff --git a/tools/testing/selftests/xcall_prefetch/echo_test.c b/tools/testing/selftests/xcall_prefetch/echo_test.c new file mode 100644 index 000000000000..fc7180869f3c --- /dev/null +++ b/tools/testing/selftests/xcall_prefetch/echo_test.c @@ -0,0 +1,52 @@ +#include <stdio.h> +#include <stdlib.h> +#include <unistd.h> +#include <sys/wait.h> + +#include "./utils.h" + +int main() { + pid_t server_pid, client_pid; + int server_status, client_status; + + server_pid = fork(); + if (server_pid == 0) { + execl("./echo_server", "./echo_server", NULL); + printf("execl echo_server failed\n"); + exit(EXIT_FAILURE); + } else if (server_pid < 0) { + printf("fork echo_server failed\n"); + exit(EXIT_FAILURE); + } + + sleep(1); + + client_pid = fork(); + if (client_pid == 0) { + execl("./echo_client", "./echo_client", NULL); + printf("execl client failed\n"); + exit(EXIT_FAILURE); + } else if (client_pid < 0) { + printf("fork echo_client failed\n"); + exit(EXIT_FAILURE); + } + + waitpid(client_pid, &client_status, 0); + + kill(server_pid, SIGTERM); + waitpid(server_pid, &server_status, 0); + + if (WIFEXITED(client_status)) { + int exit_code = WEXITSTATUS(client_status); + if (exit_code == 0) { + printf("echo test %sok%s.\n", ANSI_COLOR_GREEN, ANSI_COLOR_RESET); + exit(EXIT_SUCCESS); + } else { + printf("echo test %sfailed%s.\n", ANSI_COLOR_RED, ANSI_COLOR_RESET); + exit(EXIT_FAILURE); + } + } else { + printf("echo test %sfailed%s.\n", ANSI_COLOR_RED, ANSI_COLOR_RESET); + exit(EXIT_FAILURE); + } +} \ No newline at end of file diff --git a/tools/testing/selftests/xcall_prefetch/mulit_close_test.c b/tools/testing/selftests/xcall_prefetch/mulit_close_test.c new file mode 100644 index 000000000000..fae0407e8473 --- /dev/null +++ b/tools/testing/selftests/xcall_prefetch/mulit_close_test.c @@ -0,0 +1,48 @@ +#include <stdio.h> +#include <stdlib.h> +#include <unistd.h> +#include <sys/epoll.h> +#include <sys/socket.h> + +#include "./utils.h" + +int main() { + int epoll_fd; + struct epoll_event ev, events[1]; + int ret = -1; + + int fds[2]; + socketpair(AF_UNIX, SOCK_STREAM, 0, fds); + + epoll_fd = epoll_create1(0); + ev.events = EPOLLIN | EPOLLHUP; + ev.data.fd = fds[0]; + epoll_ctl(epoll_fd, EPOLL_CTL_ADD, fds[0], &ev); + + if (fork() == 0) { + close(fds[1]); + exit(0); + } + + close(fds[1]); + + int nfds = epoll_wait(epoll_fd, events, 1, 1000); + + if (nfds == 1) { + if (events[0].events & EPOLLHUP) { + ret = 0; + } + } + + close(fds[0]); + close(epoll_fd); + if (!ret) { + debug_printf("epoll_wait detected EPOLLHUP!\n"); + printf("multi close test %sok%s.\n", ANSI_COLOR_GREEN, ANSI_COLOR_RESET); + return 0; + } + + printf("multi close test %sfailed%s.\n", ANSI_COLOR_RED, ANSI_COLOR_RESET); + + return -1; +} \ No newline at end of file diff --git a/tools/testing/selftests/xcall_prefetch/multi_write_test.c b/tools/testing/selftests/xcall_prefetch/multi_write_test.c new file mode 100644 index 000000000000..0af07b742e73 --- /dev/null +++ b/tools/testing/selftests/xcall_prefetch/multi_write_test.c @@ -0,0 +1,77 @@ +#include <stdio.h> +#include <stdlib.h> +#include <unistd.h> +#include <sys/epoll.h> +#include <sys/wait.h> + +#include "./utils.h" + +#define NUM_CHILDREN 3 + +int main() { + int epoll_fd, pipe_fd[2]; + struct epoll_event ev, events[MAX_EVENTS]; + char buffer[BUFFER_SIZE]; + + if (pipe(pipe_fd) == -1) { + handle_error("pipe failed\n"); + } + + epoll_fd = epoll_create1(0); + if (epoll_fd == -1) { + handle_error("epoll_create1 failed\n"); + } + + ev.events = EPOLLIN; + ev.data.fd = pipe_fd[0]; + if (epoll_ctl(epoll_fd, EPOLL_CTL_ADD, pipe_fd[0], &ev) == -1) { + handle_error("epoll_ctl failed\n"); + } + + for (int i = 0; i < NUM_CHILDREN; i++) { + sleep(1); + pid_t pid = fork(); + if (pid == 0) { + close(pipe_fd[0]); + char msg[10]; + sprintf(msg, "%d %d\n", 2*i, 2*i+1); + write(pipe_fd[1], msg, strlen(msg)); + exit(0); + } else if (pid < 0) { + handle_error("fork failed\n"); + } + } + + sleep(1); + + int nfds = epoll_wait(epoll_fd, events, MAX_EVENTS, -1); + if (nfds == -1) { + handle_error("epoll_wait failed\n"); + } + + debug_printf("epoll_wait returned %d events\n", nfds); + + for (int i = 0; i < nfds; i++) { + if (events[i].data.fd == pipe_fd[0]) { + ssize_t n = read(pipe_fd[0], buffer, BUFFER_SIZE); + if (n > 0) { + debug_printf("Received: %.*s", (int)n, buffer); + } + char *expected_msg = "0 1\n2 3\n4 5\n"; + if (strncmp(expected_msg, buffer, strlen(expected_msg)) != 0) { + printf("multi write test %sfailed%s.\n", ANSI_COLOR_RED, ANSI_COLOR_RESET); + handle_error("expected msg: %s, buffer: %s", expected_msg, buffer); + } + } + } + + for (int i = 0; i < NUM_CHILDREN; i++) { + wait(NULL); + } + + close(pipe_fd[0]); + close(pipe_fd[1]); + close(epoll_fd); + printf("multi write test %sok%s.\n", ANSI_COLOR_GREEN, ANSI_COLOR_RESET); + return 0; +} \ No newline at end of file diff --git a/tools/testing/selftests/xcall_prefetch/signal_recovery_test.c b/tools/testing/selftests/xcall_prefetch/signal_recovery_test.c new file mode 100644 index 000000000000..9645f8ffec1b --- /dev/null +++ b/tools/testing/selftests/xcall_prefetch/signal_recovery_test.c @@ -0,0 +1,164 @@ +#include <stdio.h> +#include <stdlib.h> +#include <unistd.h> +#include <signal.h> +#include <sys/epoll.h> +#include <sys/socket.h> +#include <netinet/in.h> +#include <errno.h> +#include <string.h> +#include <pthread.h> + +#include "./utils.h" + + +int epoll_fd; +int server_fd; +volatile sig_atomic_t signal_received = 0; + +void signal_handler() { + signal_received = 1; +} + +void* worker_thread() { + struct epoll_event events[MAX_EVENTS]; + + while (!signal_received) { + int nfds = epoll_wait(epoll_fd, events, MAX_EVENTS, -1); + if (nfds == -1) { + if (errno == EINTR) { + debug_printf("epoll_wait interrupted by signal! re-waiting...\n"); + continue; + } else { + handle_error("epoll_wait failed\n"); + } + } + + for (int i = 0; i < nfds; i++) { + if (events[i].data.fd == server_fd) { + struct sockaddr_in address; + socklen_t addr_len = sizeof(address); + int client_fd = accept(server_fd, (struct sockaddr*)&address, &addr_len); + if (client_fd < 0) { + handle_error("accept failed\n"); + } + + struct epoll_event ev; + ev.events = EPOLLIN | EPOLLET; + ev.data.fd = client_fd; + if (epoll_ctl(epoll_fd, EPOLL_CTL_ADD, client_fd, &ev) < 0) { + close(client_fd); + handle_error("epoll_ctl failed\n"); + } + } else { + char buffer[1024]; + ssize_t n = read(events[i].data.fd, buffer, sizeof(buffer)); + if (n <= 0) { + if (n < 0) { + handle_error("read failed\n"); + } + epoll_ctl(epoll_fd, EPOLL_CTL_DEL, events[i].data.fd, NULL); + close(events[i].data.fd); + debug_printf("closing fd %d\n", events[i].data.fd); + } else { + buffer[n] = '\0'; + debug_printf("receive: %s\n", buffer); + } + } + } + } + + return NULL; +} + +void* client_thread() { + sleep(1); + + int client_fd = socket(AF_INET, SOCK_STREAM, 0); + struct sockaddr_in server_addr = { + .sin_family = AF_INET, + .sin_port = htons(8082), + .sin_addr.s_addr = htonl(INADDR_LOOPBACK), + }; + + if (connect(client_fd, (struct sockaddr*)&server_addr, sizeof(server_addr)) < 0) { + debug_printf("connect failed\n"); + goto fail; + } + + const char *msg = "hello world."; + write(client_fd, msg, strlen(msg)); + close(client_fd); + + sleep(1); + + debug_printf("Sending SIGUSR1 to interrupt epoll_wait...\n"); + kill(getpid(), SIGUSR1); + + sleep(1); + debug_printf("Testing if epoll_wait can still work after signal...\n"); + + client_fd = socket(AF_INET, SOCK_STREAM, 0); + if (connect(client_fd, (struct sockaddr*)&server_addr, sizeof(server_addr)) < 0) { + debug_printf("connect failed\n"); + goto fail; + } + + const char *msg2 = "hello world again."; + if (write(client_fd, msg2, strlen(msg2)) < 0) { + goto fail; + } + + close(client_fd); + return NULL; + +fail: + close(client_fd); + pthread_exit((void *)42); +} + +int main() { + pthread_t worker, client; + + signal(SIGUSR1, signal_handler); + + server_fd = socket(AF_INET, SOCK_STREAM, 0); + struct sockaddr_in addr = { + .sin_family = AF_INET, + .sin_port = htons(8082), + .sin_addr.s_addr = INADDR_ANY, + }; + bind(server_fd, (struct sockaddr*)&addr, sizeof(addr)); + listen(server_fd, SOMAXCONN); + + epoll_fd = epoll_create1(0); + if (epoll_fd < 0) { + handle_error("epoll_create1 failed\n"); + } + + struct epoll_event ev; + ev.events = EPOLLIN | EPOLLET; + ev.data.fd = server_fd; + if (epoll_ctl(epoll_fd, EPOLL_CTL_ADD, server_fd, &ev) < 0) { + handle_error("epoll_ctl failed\n"); + } + + pthread_create(&worker, NULL, worker_thread, NULL); + pthread_create(&client, NULL, client_thread, NULL); + + void *retval; + pthread_join(worker, NULL); + pthread_join(client, &retval); + + close(server_fd); + close(epoll_fd); + + if (retval != NULL) { + printf("signal recovery test %sfailed%s.\n", ANSI_COLOR_RED, ANSI_COLOR_RESET); + return -1; + } + + printf("signal recovery test %sok%s.\n", ANSI_COLOR_GREEN, ANSI_COLOR_RESET); + + return 0; +} \ No newline at end of file diff --git a/tools/testing/selftests/xcall_prefetch/thundering_herd_test b/tools/testing/selftests/xcall_prefetch/thundering_herd_test new file mode 100755 index 0000000000000000000000000000000000000000..8121561474f368c05f8108cddd9d80ac8b05379d GIT binary patch literal 21000 zcmeHPdvsgHnIB2Mu@pb#R}wo3M1+JU1j|n1gaA!&;)uwQ7l{LbLT_a0%2p#wM!L#_ zZ6ILWqT_lScS}oKprL15y1TFqr}P0mfYTHb*gnFZ)0Vch%@)GiOWFn#T9*d2zi;l$ z_0`3<-Lu_)_8d=+_09MD9`n32b7$`4&d|nd+%6ZB!Nb-t;&#;<NUIQd+eHMVm91hG z_+7v*X4AlzaLkam8U#`!yzDNfUV(c+$*!IvlhC;a%p{dLBuaKp;Z&I+k(5;#JlQD} z6}+83<}>spm2xw^fFWl#x=55yQmY-+&G==8-S`KKsVsLulwaDBJF@dO8v5SV#q=y? z2tiU=pHwIGw@Uc4(`GV4kkoG1E$r-cy<x^Ar8FpPKds09H|cE^c8><Q8)l~+2FxUt z>0J*ys+a$1kj=bJr03KQUQupIMfI|#WTJCrOH(q|m`tQ{`x^JHTG_a=C74YIm+>r6 zz5@88KDA-<4u*CcCQLS``=G%~e<Z6Da?=0!gEJm~;B#M_^L+2#{B4!rJ@)mBKOZF> z$~W1N4h8Zjx)|pvej)xy$Ec6HI8GRam$V=D7_rHJ7Xq%CMD!&Gd>#@COrpQpL4TJ6 zKFa}r-T~k1fdAA1zs&)^*#TeUfM4ZMUTT-g>h04G_yGqz>44wsfPcXO4?E!79q_L^ z;1@XH*E-Zg+(G}_4tSpf{z(V?at9pQVe|3#zM!hneGxUDNJWx~+cnk`NhZ@#7LTU% zB#Wj~DJ`lqRmIaBRkL~|qpLlUM2cn8(cPNPwBB?wsYaoi?e2-_tT(4;DcS{DG?~t7 z#Kd%#O=?;%>rA9#tXof~lKeL-%=bnT(CF?VH}Q04H%lh6x|U**XjJQ^#I;N&l~$3h zh@MENn6?l4BIS%0(Y2+}LEfUf)oAx_H6BSMS#Kth(&O;d8Kp62CEKvEeO;TnEO><( zTaJ-PqAg}@S#X6JU*5vh_Kr<zOv`9pl&4I`rnY1{rFBF)lNu83>Pe?Wt*D0Tq>3~G zc`)L5_`ecCJsA58OfeBdgV1RvX0(_K+P^|nj>`>+L<NmtBLd4bBrpRjY)tTVTwcQ7 z6MSwq$vw=AjzB)182Su`KR2tRj)5TOS2?e>(qKzL&YuQn48n+eTMa<!OaA4n4V=bS z;hF=dEVvxwNibr;Q6Mv%wczz8ihNpdIcAf%O5ml4TgUr)3r_Qc3;_$y*EK@0$bwTD zWN5bFl!gqeEV!%-lC@fJu{7c)VGBN8Isx5k!Nn4cOLtoEa*O^h3tnl#yDhlig7;c* z*@sBH&w^K5^bcBa>$rWd1s6+PZgt3l%dSI$`z`nki~qwGe5M8ess*PxPlh8Fyg`DX z-?rehE%>knpJTyKSa7*ECh<uNPGkC1n5u!P8knkqzg7cp`7eAY+;_$s9#DSP#8|k0 zSa%mrhWozfJ;qH7D}EkiY^<>G7a;z5jP$!nGB)}jg+k#F=V_uG8-2;-X#yJ?J!<ka zVUCSHZt^sdjg3BR@-%^sjSiVSO<ZH6cbPm*SYxBNnLJHYW1|t1rwM9o^m>!0iD_(f zt;y4bG&Z`_<Y^)r8@<TnY2qFmtu}d@kj6$`CQlR5*yx+@$@0<!gz}p_O*|;S$<u^` z@|!$OG$_Bx(*%R^n><Y{D8I?mgo5&$JWV7hzsb{r0_B%HTBnh>&U(h%2mFdgYOM*r z(Z(;d@fX<m88*Jc#w#}dU2Jn$(|g;-|JKHzw(-BT@jtflKd|x7+4!ez{5Nd;V>bTF zHvT~yf1iyXwDJ8m{(z0oS^1^EZO^~-;c)&p;l9_-ZtVyy9bWoEcyRR@JSK%1+o5?b z9`w(<2aSpcaB%g0AmKqJL*$BaeFh%Ql@!OL=_{P_&pSwZ#{`9bJJ(;aj>zKo!uhk| zqrba4e01Csc0C_{={>y;4iw?QTR0Wx?Ii8#`8~Lr#4NXDN4RhGZ<Zh$&Y#xH!vm{d z2Q%`ycMFA)7#@b_m0yB)-3nVXeX<|j3n4{zz%G2(nPKkX=`hf6{u?_vxo;OIgWVt- z^ZmV?4DJKDXIKyT7l#h|7jHV`U%cb6fAOtH{EM|=|KgsL{>8ad{>2B*h6nl$Z_lH= z*W+=f*G)E(M$Vq4!DnP4o&_@peP;ss184dUoMCs)g#H5PkDNPSC~%|E&q9A7bf$0E z6}~^5KR)siI0@%N<Keyo<1AMTgK9E3CJo+591oqN<lb06u;W}f`rUA1xGNryhmVc~ zj*h$YKRNo|yv1kpNAo9+o^rcRxQ6pDEe10fZXO6h{bYCmzFjBtFWy_#7dqzxcva}k zN1s3zCWOzU%AxkcF-sYmW8s04g^Q3d<!<S4_;~0PWdn4CXdJwoJ$q;#GJ69u8_$P+ zLzQvoNH@5A=(VRv>lvz^XK6qliJ+kQ&@17=O{WGzuRKMASUCSf)WA6OfrJNkjK{)D z>Uf&EGV<o13x%Q_QLQ4|hF==f^T&;L_!cztq4NWw^9R50&3_j<&(IJr#rH#e$l~d) zGgTPz-&ef;X6V5TCG1}%qVeM#mE}wUj_3QAfjsq1<bWshEZL0IQlb0Lzk{OiUn7?! zLJ41oK{(%EPh^0T<0N7BDq6#xFsKdm(=SZVqi2N&2Ptm`W~n&U1fQXoQfM~<Io^L3 zxghtW^maP(GhqG0{(gF&H1hfgkbHkReo@o~Af$Y}|1vX17_<+HT=Xv@)xkl62ktYn zN7lS?k?LoWUU={utt94dpBzEuJQagiNdF1ae18>$eNWSGAy-YK1}0IX_101I^WqlO z*_XIk9XF$2wr{qmyR&`FHPBBOyq7`kjp~tm{sbqiZ@dg0XCK9RJ|KSsj8W>lpxMXk z#?|*URgusC_rFH%8c%Q^03$!t9rOJqa6iC*;p8iDa=iZ>RE(M$7qK@M8Ls2t`-k_J zkGwl>WMjJC(8u*%ef^}26p8rm_|Y4#L!uZ1EZ^@%DuSi4;8$e+{_gW}ULesSW`pT_ znhagJOTvRQYOy*A53Z=@WOWsYQH+0wFCKei0knRLH4U`<hE}Di^(bkfO^}OC^z@a> zr!Nb^3&;3;&evz=ii~PtDooYDR1Hkkz*G%P)xh6Y11_=KyfDkAr*b`=S|*Uyarn>` zNNTAry*rRd?ZL@P%q}&~Spsxep~Wl*Ap)7s=6tBR2S3Y~^)S6Vm%<qYIqlXmv4E~+ z^}xbxdiTO?kQA3Mb<{EZb3MFy%<){c6~(33ef{@%a{#S-vru>xbOGr1K%W>Z6q2A1 zzEvo^0y^(pp)d~mDrh~HRJpecg;lf=1KkNK_j+BoZ)2`~RjxVZrQSmVr~TDd_`3v@ ztuLhoGroCXtOb7~zbh0DLyuKfT~k?qt-oxq_aM7^_LY||Uoami_j?QIuCs*#HzW{& z($n#`6F8+KLp>s&z~7Td?`{G^l~o^ew^h(q2<79e_?w3^Q5j8{`#Z?3AwL25804QS zlIKnN2=EoCgKLZAYfU+wtEiB-(7Ilx?OEdC3S=`CrfOiS2BvCYss^TNV5$bDYGA4c zrfT5-aSfQ?2Z<K|xfG{wlqi%N#Qb2XLhy&h#~bn+4*A`Y{6+)IaUNz1D!)IXH9Q3= zfBU^cn)oBM-bRq$@C?&}A3+fIvXIMfZDJy$IHKTz-Ww=bzfB=tn6dFS#7%rtw3Chk z5Ja&VHX?bD->TdzKK7CL36Wu$?_r@YzX_7*%Ws*a9=1n$`H)=>VV{@t-9iC7o>I;| z;&q7g3xph-teltS5Xt_Z5&2!U-C?tMS!)$^tDw6C?G^N(poauKEa(wIhXp+;=qW+Z z3d+RCN>zde1Z@_yRnV=1?h>?D(1U^=67;a3M+CK(>#y<NRzqV$Tica^MLRlkDLofh z+8kUFY;Ihd<LuHqmMsf5w=5Pgy9^daZ!+4fUG%9NWM8(Wxe9m*$GomZ=ojTj!2AOm zcrrt`xAc$DahH^y2l05_F7Nk=cbBe1nr_#$A979!z(>jL`0*^7yB<U-QSjv?E9oZa z(wQwFZm%)|o05b0@iflC$u%pv1<888(x4$e&HErJxu*RX9!q)LtGr5ycuU6sm3)*8 z8tQH$@)7!(U6&+sH#wbCcPHZB82pu%P-@D>(`aA8*N?c%_bTPm_f2x+`!67^z9(UI za5}Fl*Ys9E%su^5A|<{zA@cf;BVINA^8^LTro$4Us_Zu7K?syB=KyOhdmP#b2U+=U zvQswt?}KthubyGbX5Up1Rm5XV*-}b{tjH~2#+0q)$9Q~Y3sbHy{}JM>gdTKdn}_mT z>4T$krNjFI82_szxPe<R|7u8-8>=XV66QSw&M57)`e`%4_%<MWF5e$u;`TiRz~lQl zsrq)1oi7jRG~aWGPcIvU4vh)Z=Yj<-ISEoamp5MJb|_YPpYUD_nZFW@_nQj&XZ|bk z>;2Yr61Z6vwbIRqR#RhEu#!E9`!1|G2CmZ06Q1Y23z&Zf2`=)JF0xYskG{XF`4*`i zprm|ndw)b~Uk2j)qe$xKWHat3d$*si#`ykJ7Y9?~e~BdLYstjRs#e43*NE~=l%gN~ zBY>JHe%u!;RO!3Lhk<-g1-$El3rZNPeFQe{1wPX744uu@E<{e;3+fF1e6ChOlJ14| zgHZDf9o|zLBmE^p|4gp7f#jD7emqx8=V$K5>Ezopbnl+pcS!$cAwQR^%~FnT75r<t zy6s}DZ-yO(^GJV2g1J{z8d*4%o0*o=fl!&D?P2a!frr6*hCaQg_Eng<TLTyKB#-Q= z9iSxFRvY#QMHboxf6pGqW<DV%idI-spaefG5**3VQ%J^Dl)`d&<#ifmvo0ma7f!oK z)a13?mTEE$rF!&l_l2{`&^>fZ?HSm*=kap7hmx!&NcCF7TjG;qdgQxfP<K7>HP{<9 zUCV3QJ+yTOxoZ&i8(G6fQk-A<S&)AdL)-Tdr{H?PuP`?IBBbb=KVN2mOBR=rfqSTX z<^qxE4#E*oWLFa}vU@1T*c`e7<eEE8j(&F=TJSyl6+~Tg=TYbK41N0!tQ}m{u7#Dg zWuD48mDR4A(kir%djTseTLTD6eninhXEIOam1S3zT~XGKPmUC!fYEsxDG*kA2_xby zad|4&lo4j=y8!wY!F(?KOy?04u#88r6e*tt{j#zN-HHl~$rDLc*4pHL$jz**g3l_K ztG$*|slEs~XaZ6bbY0yrp+v=Mt^drr22!eRv6#Ejb?VAc12v^2n*o`bG|6TfGR9_E zJ$|5}$YVV|QZKS<C{}K-EmEHibu<5z$edzzQyIIjyl3yHwqn1hhI9S7vI4&}a?p{# zE6AF1*-TR+6;0-1S`%&(sViH=wW6+6uBkJDO9y{hCA;6Ik<#=8{;;2|Cz9D<HzhQ& z&92fmyGq;a!6@}AeSfb;O2}hYk70gn%;6_xAp9l*qUeprtq7IWg*L89L}Ibb1gGgn zJVDlYh8RXGpIi1}&s>+HxNoiXeo=9KN%5Xmd|gV_ca%#XR_4SN?@tw9j@Zh*ilQqQ zVsiI1DGfy2m)0nLitIG_6G|;{L1jLXMnwT}U8|IROsPAel&?`-*Oj=HfajLrQ%e15 zWrp`=rTzhB?k1)F3rgiWMe%ki<=&eV*J|ZLh^kKmc}#KPuYApGO8x6Fc$o}pZdU@{ zPbszD7nJH3p)>RXR4Tpm73CqN;wGhhr&8ai1m03A+)HLE_1<47RT!p~>LKL<?<}Rt z`yHkFQ;$*lvvg$%WpJlbamx$J2M~38Iv+AJt$XI5P-Z-=C~qmg7nRB{a^Fo#`9sP) zqfB#&=)Cmq4;jY2!c?jF*M*AmpQlkA?~96W2x+1U_i%!9n-bT}(_V9zqN`=JY%VG9 zl*A%>MAcbn>z0ihZ``(d%eGCpoRZGyILQ&{rtKRH@s`b@?cpsQQg_|j&Ffj0meMk~ zl%g7^LAV}52RFD^gIhDIO8sW}GPtKpFF5!ykfkeSg6*5x_APDKscjp#gm#A74A0xv zhj(mc*+dF=TY3`7{Z2iVXPK+wLJ!x3TimFjn>pGZErrK~t881(OT`mfzhT>26cE0t zpBnPe&bEy^wzuC9VtP-nikm>pILo3-b%C~}mx~2#ps`2KrL@(&#EB@_NG967+PLW# zXzU6!#v|GNsc3g5ol57jjk(lbTmWj+>GDh#LVjhZu{V>}amOf~X+%w>al|DB(PSjs zEycL^13@%y;f=g*+0h{`3aQaF7TqkD$|kx}S}YLlj%19=+_rv`p}cilJBueHU0H@( zJUF4FA5)*2*SJ;0>xyNK8sW9HdCTT?8__N-tLbVgo$AD8q}|l4*@#O0Mf8v&U)*qe z==!Z|+pY_B&;=(I$ACr(+(y(n+Q{eyD_amv#ba6UmXSo1VeO6LUX+SymNz4uiss}U zkEbUSgYVn5Oqw@rFFLQKz-BiNZQF)RS_$+EatRAkL>5pc(a*XgsaR5@`&a3V$T>8l z{L+;wOvN2Bx|n5@xMN$0dXyOXP?p`=et6|sRYNy~Hg{|{m>bt#7ovVGG*vtx+{5~| zwymtGJKdu-C7YX9wk%uKvZ^Va+`VFHbCaG<C$mj7GNbmJvRX1u@odw+DDJJPy%{a8 z=}~%jaSAl4l(0AXcTiRxdLMT|@(B#%6axGgGb!j~`t-hzg)}=v3YHK+obX{AzFgqu z#(}VD2X(R8tg6)jD1Q`a(c@xRw$Uot66j@${G0nPLUu2x3%eL{KSi26g2-g~cPiI2 z+)o`dPGO}NgHGn>`&@q(v+oD|n48p^_mick5w2fv?yE@rJw)iUZ+n=73YiS2Gw6U_ z8GH@q=m5j+QfslW!@-Z{fbVs{^T4a%U$&P_h0daD+|^9(Z%O<!4t^eYz>hiLKXJfE z9PoD>@G8^~<=<XE7w~$Mxt^ex0k3v9nD_mp<O;&6Wda65OBo7b;FGo2haB(}@GA8G z`c`gk{?J=%4J7jaQMg1%vO?cp|91*~`#a-395>yXkuNylj|xBbH<n`#`aco+_BX88 z1#W+HDmeJ5!MHP7y)AISFL%Jh4mfStP&?Y|ts8j2td)t#-41>}?SMb(fIkg<{)AgI zM*Ffg?s97UqVOvRKfmXAmATh-x3Hp@@5$<oZI5R3rNK1jpGapy)gxV)xAd%@i^s91 zpSaPh>OGjraHp4M-B?=fN~Sv_Ni~KA2^IjkeYgYN)0@<=Rtq*yu0eOF6KW)riR@P~ zmuB{}cqY=Lsj*y7&wiL#I81Q~y-3yJokLZx!Llr*hBmKPakF^+O`F$lYHx!xtiaSz zSV+U`w=s3Y#x3jCZdA8ibItZphuX1r-Nq0s>=(9M<xOY0(`>s?ZOZ8G^Iy8~Ot-s> zl=<cEB81-?FG5s}FLrn>A?f0K+NNfUaU<O1DJ#Yu?jnd^9xp=pE%PEo+zi)~CQ3KG z(YUm(=U3Rl6yMxdW7)K7EOt-<TcCxPJT;fa0^Rsd%XCHgGOni=L1=hd%Xen8A_czO zH*!KNeN&mTLf7C;fE+r!G8k9li!|-aPsgR0ad*DhmTu_V5nR+Sg6M925g?X#MVPAM z9=z<EbU}X7TkrhNf03PN8RO2rDY=5b3z;nMghl@+L9vJ)A^zV46v0I+Rb%m7Bxk|w z{vNz}fa)28cFPzR#9F481ygtt31UMbh*y$cEu-(ZK%Kb+{wou#-wmy`>)P=i)5W-Y zH{MfNFt$I1Bn*lzWG>mGWwNv|wlXSY87&zh0}<=RyJwJ>D~M>YD-D(+SuM(fy0#B2 zjWN(p^Jgijb&FA@8?S68XZSFN7sEovkrZAB;mj~cH)TPLYdsjQfN7n%E@(zlT^Ixf zn@GjeG9Kyd%wYdgNbujeXflo(EE>H328Pc6uw=p7)(Et@M?q5T;S~kawT6qq&{n*( zUnT4%ZMG^Sj!nIyuo9frp5{dgk}fb<^H;8SBtYwUtG#@WkTf6_1-0AX2>cH?vy=Ik z=bDnr^yRsxJ^vBdV^h#9|H~r(lF|td1<E7UowVnhZ{QR7BNLgvJV%vOo|{622YdbB z29(YurM*0dlyr@-llhl=lHLJ1opDNDo^wjNOW51<FJMN;B2=f+{)8x?q{Bi{rf)Al zZ90?vI-w`eaV3@KzGP2zX19MB7}bgNZ?0>^`bM5_Lxl%>`#%8GY9A1Gk{-5IxLh_# z`iRYbSe#=@D)o%iCzf3jKVq{#DeNV+erroUq@AQs+U(`JUDB(~iXrO>dxJABc3I0W z-=ie0kcxuZ>+d<>WG~yF1%#rY|14NZWcx|}2Z&hh9}O51Njs#Xp!W2C0o-cOE)t4@ z(t)it*wa4^AzhmmkFEJ0FaO^yyT0TleG|5Ri@mvT&<GGg>KWVwFY$LFpeB;(%lCQt z|EQf6`KRZPg0z$E|1ZEuMcT{nCdP5GngV?vZ4J_1(rJ)e?OChPZZ@b)NBWR<k}rob zVKROBo>L|4?fSOtuxgRMP?v6*v|nN(Mnu|6I$PM=)Yw(Rq1~^vllqGt>~{;p58^n< z8tnEf9PA%iVOY_@pf%W7*5+UzzTB`nU_Mv|Z@1qf?Cs@z?g~TyCD||{6Vl#FZAoy= z`{n+Z7(eC%rT0DAo^t-8{X1I{hJ70+%x`;OWx9rx_A(~7e`JH9akI_9#<KqgNsJ-j literal 0 HcmV?d00001 diff --git a/tools/testing/selftests/xcall_prefetch/thundering_herd_test.c b/tools/testing/selftests/xcall_prefetch/thundering_herd_test.c new file mode 100644 index 000000000000..bf98fcf94b6a --- /dev/null +++ b/tools/testing/selftests/xcall_prefetch/thundering_herd_test.c @@ -0,0 +1,119 @@ +#include <stdio.h> +#include <stdlib.h> +#include <unistd.h> +#include <sys/epoll.h> +#include <sys/socket.h> +#include <netinet/in.h> +#include <sys/wait.h> +#include <string.h> +#include <sys/ipc.h> +#include <sys/shm.h> +#include <fcntl.h> +#include <errno.h> + +#include "./utils.h" + +#define TIMEOUT_MS 1000 + +int create_client_connection() { + struct sockaddr_in addr = { + .sin_family = AF_INET, + .sin_port = htons(8081), + .sin_addr.s_addr = htonl(INADDR_LOOPBACK), + }; + + int fd = socket(AF_INET, SOCK_STREAM, 0); + connect(fd, (struct sockaddr *)&addr, sizeof(addr)); + return fd; +} + +void set_nonblocking(int fd) { + int flags = fcntl(fd, F_GETFL, 0); + fcntl(fd, F_SETFL, flags | O_NONBLOCK); +} + +int main() { + int listen_fd, epoll_fd; + struct sockaddr_in addr = { + .sin_family = AF_INET, + .sin_port = htons(8081), + .sin_addr.s_addr = INADDR_ANY, + }; + + int shm_id = shmget(IPC_PRIVATE, sizeof(int), IPC_CREAT | 0666); + int tmp_cnt; + int *accept_count = (int *)shmat(shm_id, NULL, 0); + *accept_count = 0; + + listen_fd = socket(AF_INET, SOCK_STREAM, 0); + set_nonblocking(listen_fd); + int reuse = 1; + setsockopt(listen_fd, SOL_SOCKET, SO_REUSEADDR, &reuse, sizeof(reuse)); + bind(listen_fd, (struct sockaddr *)&addr, sizeof(addr)); + listen(listen_fd, SOMAXCONN); + + for (int i = 0; i < NUM_THREADS; i++) { + if (fork() == 0) { + usleep(1000 * i); + epoll_fd = epoll_create1(0); + struct epoll_event ev = { + .events = EPOLLIN | EPOLLET, + .data.fd = listen_fd, + }; + epoll_ctl(epoll_fd, EPOLL_CTL_ADD, listen_fd, &ev); + + debug_printf("Worker %d waiting...\n", getpid()); + + struct epoll_event events[MAX_EVENTS]; + int nfds = epoll_wait(epoll_fd, events, MAX_EVENTS, TIMEOUT_MS); + + if (nfds == -1) { + handle_error("epoll_wait failed\n"); + } + + debug_printf("Worker %d: epoll_wait returned %d events\n", getpid(), nfds); + + int client_fd = accept(listen_fd, NULL, NULL); + if (client_fd >= 0) { + debug_printf("Worker %d accepted a connection!\n", getpid()); + __sync_fetch_and_add(accept_count, 1); + close(client_fd); + } else if (errno != EAGAIN && errno != EWOULDBLOCK) { + debug_printf("accept failed\n"); + } + + close(epoll_fd); + exit(0); + } + } + + sleep(1); + + debug_printf("Parent creating client connection...\n"); + int client_fd = create_client_connection(); + sleep(1); + close(client_fd); + + int waited = 0; + for (int i = 0; i < NUM_THREADS; i++) { + if (wait(NULL) < 0 && errno == ECHILD) { + break; + } + waited++; + } + + tmp_cnt = *accept_count; + debug_printf("Total successful accepts: %d\n", tmp_cnt); + + shmdt(accept_count); + shmctl(shm_id, IPC_RMID, NULL); + close(listen_fd); + + if (tmp_cnt == 1) { + printf("thundering herd test %sok%s.\n", ANSI_COLOR_GREEN, ANSI_COLOR_RESET); + return 0; + } else { + printf("thundering herd test %sfailed%s.\n", ANSI_COLOR_RED, ANSI_COLOR_RESET); + return -1; + } +} \ No newline at end of file diff --git a/tools/testing/selftests/xcall_prefetch/utils.h b/tools/testing/selftests/xcall_prefetch/utils.h new file mode 100644 index 000000000000..58959c564119 --- /dev/null +++ b/tools/testing/selftests/xcall_prefetch/utils.h @@ -0,0 +1,54 @@ +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <netinet/in.h> + +#define ANSI_COLOR_GREEN "\x1b[32m" +#define ANSI_COLOR_RED "\x1b[31m" +#define ANSI_COLOR_RESET "\x1b[0m" + +#define PORT 8080 +#define BUFFER_SIZE 4097 +#define MAX_EVENTS 100 +#define MAX_MSGS 4 +#define NUM_THREADS 4 + +#ifdef DEBUG +#define debug_printf(fmt, ...) (printf(fmt, ##__VA_ARGS__)) +#else +#define debug_printf(fmt, ...) (void)fmt +#endif + +struct client_info { + int fd; + int id; + struct sockaddr_in addr; +}; + +void handle_error(const char *format, ...) +{ + printf("%s", format); + exit(EXIT_FAILURE); +} + +char* generate_number_string(int length) +{ + if (length <= 0) { + printf("\nnumber string length invalid\n"); + return NULL; + } + + char* result = (char*)malloc((length + 1) * sizeof(char)); + if (result == NULL) { + printf("\nnumber string malloc failed\n"); + return NULL; + } + + for (int i = 0; i < length; i++) { + result[i] = '0' + (i % 10); + } + + result[length] = '\0'; + + return result; +} \ No newline at end of file diff --git a/tools/testing/selftests/xcall_prefetch/xcall_prefetch_test.sh b/tools/testing/selftests/xcall_prefetch/xcall_prefetch_test.sh new file mode 100755 index 000000000000..e9a990dac69a --- /dev/null +++ b/tools/testing/selftests/xcall_prefetch/xcall_prefetch_test.sh @@ -0,0 +1,39 @@ +#!/bin/bash + +RED="\033[31m" +GREEN="\033[32m" +RESET="\033[0m" + +make clean +make + +if [ -e "/proc/$$/xcall" ]; then + echo 22 > "/proc/$$/xcall" + echo @22 > "/proc/$$/xcall" + echo "enable xcall epoll prefetch, pid is $$." +else + echo "Warnning: not enable xcall." +fi + +./echo_test +sleep 1 + +./multi_write_test +sleep 1 + +./thundering_herd_test +sleep 1 + +./mulit_close_test +sleep 1 + +./signal_recovery_test +sleep 1 + +if [ $? -eq 0 ]; then + echo -e "${GREEN}tests passed successfully.${RESET}" + exit 0 +else + echo -e "${RED}tests failed.${RESET}" + exit 1 +fi \ No newline at end of file -- 2.34.1

2 1

[PATCH OLK-6.6 0/6] OLK-6.6 Bperf
by Xiaomeng Zhang 09 Jun '25

09 Jun '25

Ian Rogers (1): perf test bpf-counters: Add test for BPF event modifier Tengda Wu (2): perf stat: Support inherit events during fork() for bperf perf test: Use sqrtloop workload to test bperf event Veronika Molnarova (1): perf test stat_bpf_counter.sh: Stabilize the test results Xiaomeng Zhang (2): perf stat: Increase perf_attr_map entries perf stat: Fix incorrect display of bperf when event count is 0 tools/lib/perf/include/perf/bpf_perf.h | 1 + tools/perf/builtin-stat.c | 1 + tools/perf/tests/shell/stat_bpf_counters.sh | 79 ++++++++++----- tools/perf/util/bpf_counter.c | 61 +++++++++--- tools/perf/util/bpf_skel/bperf_follower.bpf.c | 98 +++++++++++++++++-- tools/perf/util/bpf_skel/bperf_u.h | 5 + tools/perf/util/target.h | 1 + 7 files changed, 198 insertions(+), 48 deletions(-) -- 2.34.1

2 7