- Kernel - mailweb.openeuler.org

[PATCH] wifi: mac80211: Fix deadlock in ieee80211_sta_ps_deliver_wakeup()
by Yue Haibing 18 Jul '24

18 Jul '24

From: Remi Pommarel <repk(a)triplefau.lt> stable inclusion from stable-v5.10.221 commit e7e916d693dcb5a297f40312600a82475f2e63bc category: bugfix bugzilla: https://gitee.com/src-openeuler/kernel/issues/IAD02L CVE: CVE-2024-40912 Reference: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id… --------------------------- [ Upstream commit 44c06bbde6443de206b30f513100b5670b23fc5e ] The ieee80211_sta_ps_deliver_wakeup() function takes sta->ps_lock to synchronizes with ieee80211_tx_h_unicast_ps_buf() which is called from softirq context. However using only spin_lock() to get sta->ps_lock in ieee80211_sta_ps_deliver_wakeup() does not prevent softirq to execute on this same CPU, to run ieee80211_tx_h_unicast_ps_buf() and try to take this same lock ending in deadlock. Below is an example of rcu stall that arises in such situation. rcu: INFO: rcu_sched self-detected stall on CPU rcu: 2-....: (42413413 ticks this GP) idle=b154/1/0x4000000000000000 softirq=1763/1765 fqs=21206996 rcu: (t=42586894 jiffies g=2057 q=362405 ncpus=4) CPU: 2 PID: 719 Comm: wpa_supplicant Tainted: G W 6.4.0-02158-g1b062f552873 #742 Hardware name: RPT (r1) (DT) pstate: 00000005 (nzcv daif -PAN -UAO -TCO -DIT -SSBS BTYPE=--) pc : queued_spin_lock_slowpath+0x58/0x2d0 lr : invoke_tx_handlers_early+0x5b4/0x5c0 sp : ffff00001ef64660 x29: ffff00001ef64660 x28: ffff000009bc1070 x27: ffff000009bc0ad8 x26: ffff000009bc0900 x25: ffff00001ef647a8 x24: 0000000000000000 x23: ffff000009bc0900 x22: ffff000009bc0900 x21: ffff00000ac0e000 x20: ffff00000a279e00 x19: ffff00001ef646e8 x18: 0000000000000000 x17: ffff800016468000 x16: ffff00001ef608c0 x15: 0010533c93f64f80 x14: 0010395c9faa3946 x13: 0000000000000000 x12: 00000000fa83b2da x11: 000000012edeceea x10: ffff0000010fbe00 x9 : 0000000000895440 x8 : 000000000010533c x7 : ffff00000ad8b740 x6 : ffff00000c350880 x5 : 0000000000000007 x4 : 0000000000000001 x3 : 0000000000000000 x2 : 0000000000000000 x1 : 0000000000000001 x0 : ffff00000ac0e0e8 Call trace: queued_spin_lock_slowpath+0x58/0x2d0 ieee80211_tx+0x80/0x12c ieee80211_tx_pending+0x110/0x278 tasklet_action_common.constprop.0+0x10c/0x144 tasklet_action+0x20/0x28 _stext+0x11c/0x284 ____do_softirq+0xc/0x14 call_on_irq_stack+0x24/0x34 do_softirq_own_stack+0x18/0x20 do_softirq+0x74/0x7c __local_bh_enable_ip+0xa0/0xa4 _ieee80211_wake_txqs+0x3b0/0x4b8 __ieee80211_wake_queue+0x12c/0x168 ieee80211_add_pending_skbs+0xec/0x138 ieee80211_sta_ps_deliver_wakeup+0x2a4/0x480 ieee80211_mps_sta_status_update.part.0+0xd8/0x11c ieee80211_mps_sta_status_update+0x18/0x24 sta_apply_parameters+0x3bc/0x4c0 ieee80211_change_station+0x1b8/0x2dc nl80211_set_station+0x444/0x49c genl_family_rcv_msg_doit.isra.0+0xa4/0xfc genl_rcv_msg+0x1b0/0x244 netlink_rcv_skb+0x38/0x10c genl_rcv+0x34/0x48 netlink_unicast+0x254/0x2bc netlink_sendmsg+0x190/0x3b4 ____sys_sendmsg+0x1e8/0x218 ___sys_sendmsg+0x68/0x8c __sys_sendmsg+0x44/0x84 __arm64_sys_sendmsg+0x20/0x28 do_el0_svc+0x6c/0xe8 el0_svc+0x14/0x48 el0t_64_sync_handler+0xb0/0xb4 el0t_64_sync+0x14c/0x150 Using spin_lock_bh()/spin_unlock_bh() instead prevents softirq to raise on the same CPU that is holding the lock. Fixes: 1d147bfa6429 ("mac80211: fix AP powersave TX vs. wakeup race") Signed-off-by: Remi Pommarel <repk(a)triplefau.lt> Link: https://msgid.link/8e36fe07d0fbc146f89196cd47a53c8a0afe84aa.1716910344.git.… Signed-off-by: Johannes Berg <johannes.berg(a)intel.com> Signed-off-by: Sasha Levin <sashal(a)kernel.org> Signed-off-by: Yue Haibing <yuehaibing(a)huawei.com> --- net/mac80211/sta_info.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/net/mac80211/sta_info.c b/net/mac80211/sta_info.c index 44bd03c6b847..f7637176d719 100644 --- a/net/mac80211/sta_info.c +++ b/net/mac80211/sta_info.c @@ -1343,7 +1343,7 @@ void ieee80211_sta_ps_deliver_wakeup(struct sta_info *sta) skb_queue_head_init(&pending); /* sync with ieee80211_tx_h_unicast_ps_buf */ - spin_lock(&sta->ps_lock); + spin_lock_bh(&sta->ps_lock); /* Send all buffered frames to the station */ for (ac = 0; ac < IEEE80211_NUM_ACS; ac++) { int count = skb_queue_len(&pending), tmp; @@ -1372,7 +1372,7 @@ void ieee80211_sta_ps_deliver_wakeup(struct sta_info *sta) */ clear_sta_flag(sta, WLAN_STA_PSPOLL); clear_sta_flag(sta, WLAN_STA_UAPSD); - spin_unlock(&sta->ps_lock); + spin_unlock_bh(&sta->ps_lock); atomic_dec(&ps->num_sta_ps); -- 2.34.1

1 0

[PATCH openEuler-1.0-LTS] wifi: mac80211: Fix deadlock in ieee80211_sta_ps_deliver_wakeup()
by Yue Haibing 18 Jul '24

18 Jul '24

From: Remi Pommarel <repk(a)triplefau.lt> stable inclusion from stable-v4.19.317 commit e51637e0c66a6f72d134d9f95daa47ea62b43c7e category: bugfix bugzilla: https://gitee.com/src-openeuler/kernel/issues/IACRSL CVE: CVE-2024-40912 Reference: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id… --------------------------- [ Upstream commit 44c06bbde6443de206b30f513100b5670b23fc5e ] The ieee80211_sta_ps_deliver_wakeup() function takes sta->ps_lock to synchronizes with ieee80211_tx_h_unicast_ps_buf() which is called from softirq context. However using only spin_lock() to get sta->ps_lock in ieee80211_sta_ps_deliver_wakeup() does not prevent softirq to execute on this same CPU, to run ieee80211_tx_h_unicast_ps_buf() and try to take this same lock ending in deadlock. Below is an example of rcu stall that arises in such situation. rcu: INFO: rcu_sched self-detected stall on CPU rcu: 2-....: (42413413 ticks this GP) idle=b154/1/0x4000000000000000 softirq=1763/1765 fqs=21206996 rcu: (t=42586894 jiffies g=2057 q=362405 ncpus=4) CPU: 2 PID: 719 Comm: wpa_supplicant Tainted: G W 6.4.0-02158-g1b062f552873 #742 Hardware name: RPT (r1) (DT) pstate: 00000005 (nzcv daif -PAN -UAO -TCO -DIT -SSBS BTYPE=--) pc : queued_spin_lock_slowpath+0x58/0x2d0 lr : invoke_tx_handlers_early+0x5b4/0x5c0 sp : ffff00001ef64660 x29: ffff00001ef64660 x28: ffff000009bc1070 x27: ffff000009bc0ad8 x26: ffff000009bc0900 x25: ffff00001ef647a8 x24: 0000000000000000 x23: ffff000009bc0900 x22: ffff000009bc0900 x21: ffff00000ac0e000 x20: ffff00000a279e00 x19: ffff00001ef646e8 x18: 0000000000000000 x17: ffff800016468000 x16: ffff00001ef608c0 x15: 0010533c93f64f80 x14: 0010395c9faa3946 x13: 0000000000000000 x12: 00000000fa83b2da x11: 000000012edeceea x10: ffff0000010fbe00 x9 : 0000000000895440 x8 : 000000000010533c x7 : ffff00000ad8b740 x6 : ffff00000c350880 x5 : 0000000000000007 x4 : 0000000000000001 x3 : 0000000000000000 x2 : 0000000000000000 x1 : 0000000000000001 x0 : ffff00000ac0e0e8 Call trace: queued_spin_lock_slowpath+0x58/0x2d0 ieee80211_tx+0x80/0x12c ieee80211_tx_pending+0x110/0x278 tasklet_action_common.constprop.0+0x10c/0x144 tasklet_action+0x20/0x28 _stext+0x11c/0x284 ____do_softirq+0xc/0x14 call_on_irq_stack+0x24/0x34 do_softirq_own_stack+0x18/0x20 do_softirq+0x74/0x7c __local_bh_enable_ip+0xa0/0xa4 _ieee80211_wake_txqs+0x3b0/0x4b8 __ieee80211_wake_queue+0x12c/0x168 ieee80211_add_pending_skbs+0xec/0x138 ieee80211_sta_ps_deliver_wakeup+0x2a4/0x480 ieee80211_mps_sta_status_update.part.0+0xd8/0x11c ieee80211_mps_sta_status_update+0x18/0x24 sta_apply_parameters+0x3bc/0x4c0 ieee80211_change_station+0x1b8/0x2dc nl80211_set_station+0x444/0x49c genl_family_rcv_msg_doit.isra.0+0xa4/0xfc genl_rcv_msg+0x1b0/0x244 netlink_rcv_skb+0x38/0x10c genl_rcv+0x34/0x48 netlink_unicast+0x254/0x2bc netlink_sendmsg+0x190/0x3b4 ____sys_sendmsg+0x1e8/0x218 ___sys_sendmsg+0x68/0x8c __sys_sendmsg+0x44/0x84 __arm64_sys_sendmsg+0x20/0x28 do_el0_svc+0x6c/0xe8 el0_svc+0x14/0x48 el0t_64_sync_handler+0xb0/0xb4 el0t_64_sync+0x14c/0x150 Using spin_lock_bh()/spin_unlock_bh() instead prevents softirq to raise on the same CPU that is holding the lock. Fixes: 1d147bfa6429 ("mac80211: fix AP powersave TX vs. wakeup race") Signed-off-by: Remi Pommarel <repk(a)triplefau.lt> Link: https://msgid.link/8e36fe07d0fbc146f89196cd47a53c8a0afe84aa.1716910344.git.… Signed-off-by: Johannes Berg <johannes.berg(a)intel.com> Signed-off-by: Sasha Levin <sashal(a)kernel.org> Signed-off-by: Yue Haibing <yuehaibing(a)huawei.com> --- net/mac80211/sta_info.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/net/mac80211/sta_info.c b/net/mac80211/sta_info.c index 7e66c2890008..c974ab0ceeef 100644 --- a/net/mac80211/sta_info.c +++ b/net/mac80211/sta_info.c @@ -1262,7 +1262,7 @@ void ieee80211_sta_ps_deliver_wakeup(struct sta_info *sta) skb_queue_head_init(&pending); /* sync with ieee80211_tx_h_unicast_ps_buf */ - spin_lock(&sta->ps_lock); + spin_lock_bh(&sta->ps_lock); /* Send all buffered frames to the station */ for (ac = 0; ac < IEEE80211_NUM_ACS; ac++) { int count = skb_queue_len(&pending), tmp; @@ -1291,7 +1291,7 @@ void ieee80211_sta_ps_deliver_wakeup(struct sta_info *sta) */ clear_sta_flag(sta, WLAN_STA_PSPOLL); clear_sta_flag(sta, WLAN_STA_UAPSD); - spin_unlock(&sta->ps_lock); + spin_unlock_bh(&sta->ps_lock); atomic_dec(&ps->num_sta_ps); -- 2.34.1

2 1

[PATCH OLK-6.6 0/1] kvm : add coda feature
by l00500167 18 Jul '24

18 Jul '24

From: JunBin Li <lijunbin4(a)huawei.com> virtCCA feature JunBin Li (1): kvm: add coda feature Signed-off-by: JunBin Li <lijunbin4(a)huawei.com> drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h | 14 ++++++++++++++ 1 file changed, 14 insertions(+) -- 2.25.1

2 2

[PATCH OLK-5.10] mips: bmips: BCM6358: make sure CBR is correctly set
by Zhang Changzhong 18 Jul '24

18 Jul '24

From: Christian Marangi <ansuelsmth(a)gmail.com> stable inclusion from stable-v5.10.221 commit 36d771ce6028b886e18a4a8956a5d23688e4e13d category: bugfix bugzilla: https://gitee.com/src-openeuler/kernel/issues/IACS75 CVE: CVE-2024-40963 Reference: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id… -------------------------------- [ Upstream commit ce5cdd3b05216b704a704f466fb4c2dff3778caf ] It was discovered that some device have CBR address set to 0 causing kernel panic when arch_sync_dma_for_cpu_all is called. This was notice in situation where the system is booted from TP1 and BMIPS_GET_CBR() returns 0 instead of a valid address and !!(read_c0_brcm_cmt_local() & (1 << 31)); not failing. The current check whether RAC flush should be disabled or not are not enough hence lets check if CBR is a valid address or not. Fixes: ab327f8acdf8 ("mips: bmips: BCM6358: disable RAC flush for TP1") Signed-off-by: Christian Marangi <ansuelsmth(a)gmail.com> Acked-by: Florian Fainelli <florian.fainelli(a)broadcom.com> Signed-off-by: Thomas Bogendoerfer <tsbogend(a)alpha.franken.de> Signed-off-by: Sasha Levin <sashal(a)kernel.org> Signed-off-by: Zhang Changzhong <zhangchangzhong(a)huawei.com> --- arch/mips/bmips/setup.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/arch/mips/bmips/setup.c b/arch/mips/bmips/setup.c index 1606308..ac75797 100644 --- a/arch/mips/bmips/setup.c +++ b/arch/mips/bmips/setup.c @@ -110,7 +110,8 @@ static void bcm6358_quirks(void) * RAC flush causes kernel panics on BCM6358 when booting from TP1 * because the bootloader is not initializing it properly. */ - bmips_rac_flush_disable = !!(read_c0_brcm_cmt_local() & (1 << 31)); + bmips_rac_flush_disable = !!(read_c0_brcm_cmt_local() & (1 << 31)) || + !!BMIPS_GET_CBR(); } static void bcm6368_quirks(void) -- 2.9.5

2 1

[PATCH OLK-6.6] mips: bmips: BCM6358: make sure CBR is correctly set
by Zhang Changzhong 18 Jul '24

18 Jul '24

From: Christian Marangi <ansuelsmth(a)gmail.com> stable inclusion from stable-v6.6.36 commit 2cd4854ef14a487bcfb76c7980675980cad27b52 category: bugfix bugzilla: https://gitee.com/src-openeuler/kernel/issues/IACS75 CVE: CVE-2024-40963 Reference: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id… -------------------------------- [ Upstream commit ce5cdd3b05216b704a704f466fb4c2dff3778caf ] It was discovered that some device have CBR address set to 0 causing kernel panic when arch_sync_dma_for_cpu_all is called. This was notice in situation where the system is booted from TP1 and BMIPS_GET_CBR() returns 0 instead of a valid address and !!(read_c0_brcm_cmt_local() & (1 << 31)); not failing. The current check whether RAC flush should be disabled or not are not enough hence lets check if CBR is a valid address or not. Fixes: ab327f8acdf8 ("mips: bmips: BCM6358: disable RAC flush for TP1") Signed-off-by: Christian Marangi <ansuelsmth(a)gmail.com> Acked-by: Florian Fainelli <florian.fainelli(a)broadcom.com> Signed-off-by: Thomas Bogendoerfer <tsbogend(a)alpha.franken.de> Signed-off-by: Sasha Levin <sashal(a)kernel.org> Signed-off-by: Zhang Changzhong <zhangchangzhong(a)huawei.com> --- arch/mips/bmips/setup.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/arch/mips/bmips/setup.c b/arch/mips/bmips/setup.c index ec180ab..66a8ba1 100644 --- a/arch/mips/bmips/setup.c +++ b/arch/mips/bmips/setup.c @@ -110,7 +110,8 @@ static void bcm6358_quirks(void) * RAC flush causes kernel panics on BCM6358 when booting from TP1 * because the bootloader is not initializing it properly. */ - bmips_rac_flush_disable = !!(read_c0_brcm_cmt_local() & (1 << 31)); + bmips_rac_flush_disable = !!(read_c0_brcm_cmt_local() & (1 << 31)) || + !!BMIPS_GET_CBR(); } static void bcm6368_quirks(void) -- 2.9.5

2 1

[PATCH OLK-6.6] fs/9p: fix uninitialized values during inode evict
by Long Li 18 Jul '24

18 Jul '24

From: Eric Van Hensbergen <ericvh(a)kernel.org> mainline inclusion from mainline-v6.9-rc1 commit 6630036b7c228f57c7893ee0403e92c2db2cd21d category: bugfix bugzilla: https://gitee.com/src-openeuler/kernel/issues/I9U997 CVE: CVE-2024-36923 Reference: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?… -------------------------------- If an iget fails due to not being able to retrieve information from the server then the inode structure is only partially initialized. When the inode gets evicted, references to uninitialized structures (like fscache cookies) were being made. This patch checks for a bad_inode before doing anything other than clearing the inode from the cache. Since the inode is bad, it shouldn't have any state associated with it that needs to be written back (and there really isn't a way to complete those anyways). Reported-by: syzbot+eb83fe1cce5833cd66a0(a)syzkaller.appspotmail.com Signed-off-by: Eric Van Hensbergen <ericvh(a)kernel.org> Conflicts: fs/9p/vfs_inode.c [Context changed since commit 80105ed2fd27 ("9p: Use netfslib read/write_iter") in v9fs_evict_inode] Signed-off-by: Long Li <leo.lilong(a)huawei.com> --- fs/9p/vfs_inode.c | 18 +++++++++++------- 1 file changed, 11 insertions(+), 7 deletions(-) diff --git a/fs/9p/vfs_inode.c b/fs/9p/vfs_inode.c index 853c63b83681..c7c5631fe825 100644 --- a/fs/9p/vfs_inode.c +++ b/fs/9p/vfs_inode.c @@ -374,20 +374,24 @@ void v9fs_evict_inode(struct inode *inode) struct v9fs_inode __maybe_unused *v9inode = V9FS_I(inode); __le32 __maybe_unused version; - truncate_inode_pages_final(&inode->i_data); + if (!is_bad_inode(inode)) { + truncate_inode_pages_final(&inode->i_data); #ifdef CONFIG_9P_FSCACHE - version = cpu_to_le32(v9inode->qid.version); - fscache_clear_inode_writeback(v9fs_inode_cookie(v9inode), inode, - &version); + version = cpu_to_le32(v9inode->qid.version); + fscache_clear_inode_writeback(v9fs_inode_cookie(v9inode), inode, + &version); #endif - clear_inode(inode); - filemap_fdatawrite(&inode->i_data); + clear_inode(inode); + filemap_fdatawrite(&inode->i_data); #ifdef CONFIG_9P_FSCACHE - fscache_relinquish_cookie(v9fs_inode_cookie(v9inode), false); + if (v9fs_inode_cookie(v9inode)) + fscache_relinquish_cookie(v9fs_inode_cookie(v9inode), false); #endif + } else + clear_inode(inode); } static int v9fs_test_inode(struct inode *inode, void *data) -- 2.39.2

2 1

[PATCH openEuler-1.0-LTS 0/2] CVE-2022-48836
by Yuan Can 18 Jul '24

18 Jul '24

Johan Hovold (1): Input: aiptek - fix endpoint sanity check Pavel Skripkin (1): Input: aiptek - properly check endpoint type drivers/input/tablet/aiptek.c | 10 ++++------ 1 file changed, 4 insertions(+), 6 deletions(-) -- 2.17.1

2 3

[PATCH openEuler-22.03-LTS-SP1] bpf: Fix overrunning reservations in ringbuf
by Pu Lehui 18 Jul '24

18 Jul '24

From: Daniel Borkmann <daniel(a)iogearbox.net> mainline inclusion from mainline-v6.10-rc6 commit cfa1a2329a691ffd991fcf7248a57d752e712881 category: bugfix bugzilla: https://gitee.com/src-openeuler/kernel/issues/IADR59 CVE: CVE-2024-41009 Reference: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?… -------------------------------- The BPF ring buffer internally is implemented as a power-of-2 sized circular buffer, with two logical and ever-increasing counters: consumer_pos is the consumer counter to show which logical position the consumer consumed the data, and producer_pos which is the producer counter denoting the amount of data reserved by all producers. Each time a record is reserved, the producer that "owns" the record will successfully advance producer counter. In user space each time a record is read, the consumer of the data advanced the consumer counter once it finished processing. Both counters are stored in separate pages so that from user space, the producer counter is read-only and the consumer counter is read-write. One aspect that simplifies and thus speeds up the implementation of both producers and consumers is how the data area is mapped twice contiguously back-to-back in the virtual memory, allowing to not take any special measures for samples that have to wrap around at the end of the circular buffer data area, because the next page after the last data page would be first data page again, and thus the sample will still appear completely contiguous in virtual memory. Each record has a struct bpf_ringbuf_hdr { u32 len; u32 pg_off; } header for book-keeping the length and offset, and is inaccessible to the BPF program. Helpers like bpf_ringbuf_reserve() return `(void *)hdr + BPF_RINGBUF_HDR_SZ` for the BPF program to use. Bing-Jhong and Muhammad reported that it is however possible to make a second allocated memory chunk overlapping with the first chunk and as a result, the BPF program is now able to edit first chunk's header. For example, consider the creation of a BPF_MAP_TYPE_RINGBUF map with size of 0x4000. Next, the consumer_pos is modified to 0x3000 /before/ a call to bpf_ringbuf_reserve() is made. This will allocate a chunk A, which is in [0x0,0x3008], and the BPF program is able to edit [0x8,0x3008]. Now, lets allocate a chunk B with size 0x3000. This will succeed because consumer_pos was edited ahead of time to pass the `new_prod_pos - cons_pos > rb->mask` check. Chunk B will be in range [0x3008,0x6010], and the BPF program is able to edit [0x3010,0x6010]. Due to the ring buffer memory layout mentioned earlier, the ranges [0x0,0x4000] and [0x4000,0x8000] point to the same data pages. This means that chunk B at [0x4000,0x4008] is chunk A's header. bpf_ringbuf_submit() / bpf_ringbuf_discard() use the header's pg_off to then locate the bpf_ringbuf itself via bpf_ringbuf_restore_from_rec(). Once chunk B modified chunk A's header, then bpf_ringbuf_commit() refers to the wrong page and could cause a crash. Fix it by calculating the oldest pending_pos and check whether the range from the oldest outstanding record to the newest would span beyond the ring buffer size. If that is the case, then reject the request. We've tested with the ring buffer benchmark in BPF selftests (./benchs/run_bench_ringbufs.sh) before/after the fix and while it seems a bit slower on some benchmarks, it is still not significantly enough to matter. Fixes: 457f44363a88 ("bpf: Implement BPF ring buffer and verifier support for it") Reported-by: Bing-Jhong Billy Jheng <billy(a)starlabs.sg> Reported-by: Muhammad Ramdhan <ramdhan(a)starlabs.sg> Co-developed-by: Bing-Jhong Billy Jheng <billy(a)starlabs.sg> Co-developed-by: Andrii Nakryiko <andrii(a)kernel.org> Signed-off-by: Bing-Jhong Billy Jheng <billy(a)starlabs.sg> Signed-off-by: Andrii Nakryiko <andrii(a)kernel.org> Signed-off-by: Daniel Borkmann <daniel(a)iogearbox.net> Signed-off-by: Andrii Nakryiko <andrii(a)kernel.org> Link: https://lore.kernel.org/bpf/20240621140828.18238-1-daniel@iogearbox.net Conflicts: kernel/bpf/ringbuf.c [The Conflicts were due to not merge commit 583c1f420173f] Signed-off-by: Pu Lehui <pulehui(a)huawei.com> --- kernel/bpf/ringbuf.c | 28 +++++++++++++++++++++++----- 1 file changed, 23 insertions(+), 5 deletions(-) diff --git a/kernel/bpf/ringbuf.c b/kernel/bpf/ringbuf.c index d6fbe17432ae..d21d4ba2eb39 100644 --- a/kernel/bpf/ringbuf.c +++ b/kernel/bpf/ringbuf.c @@ -44,6 +44,7 @@ struct bpf_ringbuf { */ unsigned long consumer_pos __aligned(PAGE_SIZE); unsigned long producer_pos __aligned(PAGE_SIZE); + unsigned long pending_pos; char data[] __aligned(PAGE_SIZE); }; @@ -145,6 +146,7 @@ static struct bpf_ringbuf *bpf_ringbuf_alloc(size_t data_sz, int numa_node) rb->mask = data_sz - 1; rb->consumer_pos = 0; rb->producer_pos = 0; + rb->pending_pos = 0; return rb; } @@ -323,9 +325,9 @@ bpf_ringbuf_restore_from_rec(struct bpf_ringbuf_hdr *hdr) static void *__bpf_ringbuf_reserve(struct bpf_ringbuf *rb, u64 size) { - unsigned long cons_pos, prod_pos, new_prod_pos, flags; - u32 len, pg_off; + unsigned long cons_pos, prod_pos, new_prod_pos, pend_pos, flags; struct bpf_ringbuf_hdr *hdr; + u32 len, pg_off, tmp_size, hdr_len; if (unlikely(size > RINGBUF_MAX_RECORD_SZ)) return NULL; @@ -343,13 +345,29 @@ static void *__bpf_ringbuf_reserve(struct bpf_ringbuf *rb, u64 size) spin_lock_irqsave(&rb->spinlock, flags); } + pend_pos = rb->pending_pos; prod_pos = rb->producer_pos; new_prod_pos = prod_pos + len; - /* check for out of ringbuf space by ensuring producer position - * doesn't advance more than (ringbuf_size - 1) ahead + while (pend_pos < prod_pos) { + hdr = (void *)rb->data + (pend_pos & rb->mask); + hdr_len = READ_ONCE(hdr->len); + if (hdr_len & BPF_RINGBUF_BUSY_BIT) + break; + tmp_size = hdr_len & ~BPF_RINGBUF_DISCARD_BIT; + tmp_size = round_up(tmp_size + BPF_RINGBUF_HDR_SZ, 8); + pend_pos += tmp_size; + } + rb->pending_pos = pend_pos; + + /* check for out of ringbuf space: + * - by ensuring producer position doesn't advance more than + * (ringbuf_size - 1) ahead + * - by ensuring oldest not yet committed record until newest + * record does not span more than (ringbuf_size - 1) */ - if (new_prod_pos - cons_pos > rb->mask) { + if (new_prod_pos - cons_pos > rb->mask || + new_prod_pos - pend_pos > rb->mask) { spin_unlock_irqrestore(&rb->spinlock, flags); return NULL; } -- 2.34.1

2 1

[PATCH OLK-5.10] bpf: Fix overrunning reservations in ringbuf
by Pu Lehui 18 Jul '24

18 Jul '24

From: Daniel Borkmann <daniel(a)iogearbox.net> mainline inclusion from mainline-v6.10-rc6 commit cfa1a2329a691ffd991fcf7248a57d752e712881 category: bugfix bugzilla: https://gitee.com/src-openeuler/kernel/issues/IADR59 CVE: CVE-2024-41009 Reference: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?… -------------------------------- The BPF ring buffer internally is implemented as a power-of-2 sized circular buffer, with two logical and ever-increasing counters: consumer_pos is the consumer counter to show which logical position the consumer consumed the data, and producer_pos which is the producer counter denoting the amount of data reserved by all producers. Each time a record is reserved, the producer that "owns" the record will successfully advance producer counter. In user space each time a record is read, the consumer of the data advanced the consumer counter once it finished processing. Both counters are stored in separate pages so that from user space, the producer counter is read-only and the consumer counter is read-write. One aspect that simplifies and thus speeds up the implementation of both producers and consumers is how the data area is mapped twice contiguously back-to-back in the virtual memory, allowing to not take any special measures for samples that have to wrap around at the end of the circular buffer data area, because the next page after the last data page would be first data page again, and thus the sample will still appear completely contiguous in virtual memory. Each record has a struct bpf_ringbuf_hdr { u32 len; u32 pg_off; } header for book-keeping the length and offset, and is inaccessible to the BPF program. Helpers like bpf_ringbuf_reserve() return `(void *)hdr + BPF_RINGBUF_HDR_SZ` for the BPF program to use. Bing-Jhong and Muhammad reported that it is however possible to make a second allocated memory chunk overlapping with the first chunk and as a result, the BPF program is now able to edit first chunk's header. For example, consider the creation of a BPF_MAP_TYPE_RINGBUF map with size of 0x4000. Next, the consumer_pos is modified to 0x3000 /before/ a call to bpf_ringbuf_reserve() is made. This will allocate a chunk A, which is in [0x0,0x3008], and the BPF program is able to edit [0x8,0x3008]. Now, lets allocate a chunk B with size 0x3000. This will succeed because consumer_pos was edited ahead of time to pass the `new_prod_pos - cons_pos > rb->mask` check. Chunk B will be in range [0x3008,0x6010], and the BPF program is able to edit [0x3010,0x6010]. Due to the ring buffer memory layout mentioned earlier, the ranges [0x0,0x4000] and [0x4000,0x8000] point to the same data pages. This means that chunk B at [0x4000,0x4008] is chunk A's header. bpf_ringbuf_submit() / bpf_ringbuf_discard() use the header's pg_off to then locate the bpf_ringbuf itself via bpf_ringbuf_restore_from_rec(). Once chunk B modified chunk A's header, then bpf_ringbuf_commit() refers to the wrong page and could cause a crash. Fix it by calculating the oldest pending_pos and check whether the range from the oldest outstanding record to the newest would span beyond the ring buffer size. If that is the case, then reject the request. We've tested with the ring buffer benchmark in BPF selftests (./benchs/run_bench_ringbufs.sh) before/after the fix and while it seems a bit slower on some benchmarks, it is still not significantly enough to matter. Fixes: 457f44363a88 ("bpf: Implement BPF ring buffer and verifier support for it") Reported-by: Bing-Jhong Billy Jheng <billy(a)starlabs.sg> Reported-by: Muhammad Ramdhan <ramdhan(a)starlabs.sg> Co-developed-by: Bing-Jhong Billy Jheng <billy(a)starlabs.sg> Co-developed-by: Andrii Nakryiko <andrii(a)kernel.org> Signed-off-by: Bing-Jhong Billy Jheng <billy(a)starlabs.sg> Signed-off-by: Andrii Nakryiko <andrii(a)kernel.org> Signed-off-by: Daniel Borkmann <daniel(a)iogearbox.net> Signed-off-by: Andrii Nakryiko <andrii(a)kernel.org> Link: https://lore.kernel.org/bpf/20240621140828.18238-1-daniel@iogearbox.net Conflicts: kernel/bpf/ringbuf.c [The Conflicts were due to not merge commit 583c1f420173f] Signed-off-by: Pu Lehui <pulehui(a)huawei.com> --- kernel/bpf/ringbuf.c | 28 +++++++++++++++++++++++----- 1 file changed, 23 insertions(+), 5 deletions(-) diff --git a/kernel/bpf/ringbuf.c b/kernel/bpf/ringbuf.c index d6fbe17432ae..d21d4ba2eb39 100644 --- a/kernel/bpf/ringbuf.c +++ b/kernel/bpf/ringbuf.c @@ -44,6 +44,7 @@ struct bpf_ringbuf { */ unsigned long consumer_pos __aligned(PAGE_SIZE); unsigned long producer_pos __aligned(PAGE_SIZE); + unsigned long pending_pos; char data[] __aligned(PAGE_SIZE); }; @@ -145,6 +146,7 @@ static struct bpf_ringbuf *bpf_ringbuf_alloc(size_t data_sz, int numa_node) rb->mask = data_sz - 1; rb->consumer_pos = 0; rb->producer_pos = 0; + rb->pending_pos = 0; return rb; } @@ -323,9 +325,9 @@ bpf_ringbuf_restore_from_rec(struct bpf_ringbuf_hdr *hdr) static void *__bpf_ringbuf_reserve(struct bpf_ringbuf *rb, u64 size) { - unsigned long cons_pos, prod_pos, new_prod_pos, flags; - u32 len, pg_off; + unsigned long cons_pos, prod_pos, new_prod_pos, pend_pos, flags; struct bpf_ringbuf_hdr *hdr; + u32 len, pg_off, tmp_size, hdr_len; if (unlikely(size > RINGBUF_MAX_RECORD_SZ)) return NULL; @@ -343,13 +345,29 @@ static void *__bpf_ringbuf_reserve(struct bpf_ringbuf *rb, u64 size) spin_lock_irqsave(&rb->spinlock, flags); } + pend_pos = rb->pending_pos; prod_pos = rb->producer_pos; new_prod_pos = prod_pos + len; - /* check for out of ringbuf space by ensuring producer position - * doesn't advance more than (ringbuf_size - 1) ahead + while (pend_pos < prod_pos) { + hdr = (void *)rb->data + (pend_pos & rb->mask); + hdr_len = READ_ONCE(hdr->len); + if (hdr_len & BPF_RINGBUF_BUSY_BIT) + break; + tmp_size = hdr_len & ~BPF_RINGBUF_DISCARD_BIT; + tmp_size = round_up(tmp_size + BPF_RINGBUF_HDR_SZ, 8); + pend_pos += tmp_size; + } + rb->pending_pos = pend_pos; + + /* check for out of ringbuf space: + * - by ensuring producer position doesn't advance more than + * (ringbuf_size - 1) ahead + * - by ensuring oldest not yet committed record until newest + * record does not span more than (ringbuf_size - 1) */ - if (new_prod_pos - cons_pos > rb->mask) { + if (new_prod_pos - cons_pos > rb->mask || + new_prod_pos - pend_pos > rb->mask) { spin_unlock_irqrestore(&rb->spinlock, flags); return NULL; } -- 2.34.1

2 1

[PATCH OLK-6.6] bpf: Fix overrunning reservations in ringbuf
by Pu Lehui 18 Jul '24

18 Jul '24

From: Daniel Borkmann <daniel(a)iogearbox.net> stable inclusion from stable-v6.6.37 commit 511804ab701c0503b72eac08217eabfd366ba069 category: bugfix bugzilla: https://gitee.com/src-openeuler/kernel/issues/IADR59 CVE: CVE-2024-41009 Reference: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id… -------------------------------- [ Upstream commit cfa1a2329a691ffd991fcf7248a57d752e712881 ] The BPF ring buffer internally is implemented as a power-of-2 sized circular buffer, with two logical and ever-increasing counters: consumer_pos is the consumer counter to show which logical position the consumer consumed the data, and producer_pos which is the producer counter denoting the amount of data reserved by all producers. Each time a record is reserved, the producer that "owns" the record will successfully advance producer counter. In user space each time a record is read, the consumer of the data advanced the consumer counter once it finished processing. Both counters are stored in separate pages so that from user space, the producer counter is read-only and the consumer counter is read-write. One aspect that simplifies and thus speeds up the implementation of both producers and consumers is how the data area is mapped twice contiguously back-to-back in the virtual memory, allowing to not take any special measures for samples that have to wrap around at the end of the circular buffer data area, because the next page after the last data page would be first data page again, and thus the sample will still appear completely contiguous in virtual memory. Each record has a struct bpf_ringbuf_hdr { u32 len; u32 pg_off; } header for book-keeping the length and offset, and is inaccessible to the BPF program. Helpers like bpf_ringbuf_reserve() return `(void *)hdr + BPF_RINGBUF_HDR_SZ` for the BPF program to use. Bing-Jhong and Muhammad reported that it is however possible to make a second allocated memory chunk overlapping with the first chunk and as a result, the BPF program is now able to edit first chunk's header. For example, consider the creation of a BPF_MAP_TYPE_RINGBUF map with size of 0x4000. Next, the consumer_pos is modified to 0x3000 /before/ a call to bpf_ringbuf_reserve() is made. This will allocate a chunk A, which is in [0x0,0x3008], and the BPF program is able to edit [0x8,0x3008]. Now, lets allocate a chunk B with size 0x3000. This will succeed because consumer_pos was edited ahead of time to pass the `new_prod_pos - cons_pos > rb->mask` check. Chunk B will be in range [0x3008,0x6010], and the BPF program is able to edit [0x3010,0x6010]. Due to the ring buffer memory layout mentioned earlier, the ranges [0x0,0x4000] and [0x4000,0x8000] point to the same data pages. This means that chunk B at [0x4000,0x4008] is chunk A's header. bpf_ringbuf_submit() / bpf_ringbuf_discard() use the header's pg_off to then locate the bpf_ringbuf itself via bpf_ringbuf_restore_from_rec(). Once chunk B modified chunk A's header, then bpf_ringbuf_commit() refers to the wrong page and could cause a crash. Fix it by calculating the oldest pending_pos and check whether the range from the oldest outstanding record to the newest would span beyond the ring buffer size. If that is the case, then reject the request. We've tested with the ring buffer benchmark in BPF selftests (./benchs/run_bench_ringbufs.sh) before/after the fix and while it seems a bit slower on some benchmarks, it is still not significantly enough to matter. Fixes: 457f44363a88 ("bpf: Implement BPF ring buffer and verifier support for it") Reported-by: Bing-Jhong Billy Jheng <billy(a)starlabs.sg> Reported-by: Muhammad Ramdhan <ramdhan(a)starlabs.sg> Co-developed-by: Bing-Jhong Billy Jheng <billy(a)starlabs.sg> Co-developed-by: Andrii Nakryiko <andrii(a)kernel.org> Signed-off-by: Bing-Jhong Billy Jheng <billy(a)starlabs.sg> Signed-off-by: Andrii Nakryiko <andrii(a)kernel.org> Signed-off-by: Daniel Borkmann <daniel(a)iogearbox.net> Signed-off-by: Andrii Nakryiko <andrii(a)kernel.org> Link: https://lore.kernel.org/bpf/20240621140828.18238-1-daniel@iogearbox.net Signed-off-by: Sasha Levin <sashal(a)kernel.org> Signed-off-by: Pu Lehui <pulehui(a)huawei.com> --- kernel/bpf/ringbuf.c | 31 +++++++++++++++++++++++++------ 1 file changed, 25 insertions(+), 6 deletions(-) diff --git a/kernel/bpf/ringbuf.c b/kernel/bpf/ringbuf.c index f045fde632e5..238d9b206bbd 100644 --- a/kernel/bpf/ringbuf.c +++ b/kernel/bpf/ringbuf.c @@ -51,7 +51,8 @@ struct bpf_ringbuf { * This prevents a user-space application from modifying the * position and ruining in-kernel tracking. The permissions of the * pages depend on who is producing samples: user-space or the - * kernel. + * kernel. Note that the pending counter is placed in the same + * page as the producer, so that it shares the same cache line. * * Kernel-producer * --------------- @@ -70,6 +71,7 @@ struct bpf_ringbuf { */ unsigned long consumer_pos __aligned(PAGE_SIZE); unsigned long producer_pos __aligned(PAGE_SIZE); + unsigned long pending_pos; char data[] __aligned(PAGE_SIZE); }; @@ -179,6 +181,7 @@ static struct bpf_ringbuf *bpf_ringbuf_alloc(size_t data_sz, int numa_node) rb->mask = data_sz - 1; rb->consumer_pos = 0; rb->producer_pos = 0; + rb->pending_pos = 0; return rb; } @@ -404,9 +407,9 @@ bpf_ringbuf_restore_from_rec(struct bpf_ringbuf_hdr *hdr) static void *__bpf_ringbuf_reserve(struct bpf_ringbuf *rb, u64 size) { - unsigned long cons_pos, prod_pos, new_prod_pos, flags; - u32 len, pg_off; + unsigned long cons_pos, prod_pos, new_prod_pos, pend_pos, flags; struct bpf_ringbuf_hdr *hdr; + u32 len, pg_off, tmp_size, hdr_len; if (unlikely(size > RINGBUF_MAX_RECORD_SZ)) return NULL; @@ -424,13 +427,29 @@ static void *__bpf_ringbuf_reserve(struct bpf_ringbuf *rb, u64 size) spin_lock_irqsave(&rb->spinlock, flags); } + pend_pos = rb->pending_pos; prod_pos = rb->producer_pos; new_prod_pos = prod_pos + len; - /* check for out of ringbuf space by ensuring producer position - * doesn't advance more than (ringbuf_size - 1) ahead + while (pend_pos < prod_pos) { + hdr = (void *)rb->data + (pend_pos & rb->mask); + hdr_len = READ_ONCE(hdr->len); + if (hdr_len & BPF_RINGBUF_BUSY_BIT) + break; + tmp_size = hdr_len & ~BPF_RINGBUF_DISCARD_BIT; + tmp_size = round_up(tmp_size + BPF_RINGBUF_HDR_SZ, 8); + pend_pos += tmp_size; + } + rb->pending_pos = pend_pos; + + /* check for out of ringbuf space: + * - by ensuring producer position doesn't advance more than + * (ringbuf_size - 1) ahead + * - by ensuring oldest not yet committed record until newest + * record does not span more than (ringbuf_size - 1) */ - if (new_prod_pos - cons_pos > rb->mask) { + if (new_prod_pos - cons_pos > rb->mask || + new_prod_pos - pend_pos > rb->mask) { spin_unlock_irqrestore(&rb->spinlock, flags); return NULL; } -- 2.34.1

2 1