- Kernel - mailweb.openeuler.org

[PATCH openEuler-22.03-LTS-SP1 0/2] CVE-2024-36933
by Ziyang Xuan 09 Jun '24

09 Jun '24

Patchset of CVE-2024-36933. Dong Chenchen (1): net: nsh: Use correct mac_offset to unwind gso skb in nsh_gso_segment() Kuniyuki Iwashima (1): nsh: Restore skb->{protocol,data,mac_header} for outer header in nsh_gso_segment(). net/nsh/nsh.c | 20 ++++++++++---------- 1 file changed, 10 insertions(+), 10 deletions(-) -- 2.25.1

2 3

[PATCH OLK-6.6 V1] scsi: lpfc: Move NPIV's transport unregistration to after resource clean up
by Cheng Yu 09 Jun '24

09 Jun '24

From: Justin Tee <justin.tee(a)broadcom.com> mainline inclusion from mainline-v6.9-rc2 commit 4ddf01f2f1504fa08b766e8cfeec558e9f8eef6c category: bugfix bugzilla: https://gitee.com/src-openeuler/kernel/issues/I9UMD0 CVE: CVE-2024-36952 Reference: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?… -------------------------------- There are cases after NPIV deletion where the fabric switch still believes the NPIV is logged into the fabric. This occurs when a vport is unregistered before the Remove All DA_ID CT and LOGO ELS are sent to the fabric. Currently fc_remove_host(), which calls dev_loss_tmo for all D_IDs including the fabric D_ID, removes the last ndlp reference and frees the ndlp rport object. This sometimes causes the race condition where the final DA_ID and LOGO are skipped from being sent to the fabric switch. Fix by moving the fc_remove_host() and scsi_remove_host() calls after DA_ID and LOGO are sent. Signed-off-by: Justin Tee <justin.tee(a)broadcom.com> Link: https://lore.kernel.org/r/20240305200503.57317-3-justintee8345@gmail.com Signed-off-by: Martin K. Petersen <martin.petersen(a)oracle.com> Signed-off-by: Cheng Yu <serein.chengyu(a)huawei.com> --- drivers/scsi/lpfc/lpfc_vport.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/drivers/scsi/lpfc/lpfc_vport.c b/drivers/scsi/lpfc/lpfc_vport.c index 6c7559cf1a4b..9e0e9e02d2c4 100644 --- a/drivers/scsi/lpfc/lpfc_vport.c +++ b/drivers/scsi/lpfc/lpfc_vport.c @@ -683,10 +683,6 @@ lpfc_vport_delete(struct fc_vport *fc_vport) lpfc_free_sysfs_attr(vport); lpfc_debugfs_terminate(vport); - /* Remove FC host to break driver binding. */ - fc_remove_host(shost); - scsi_remove_host(shost); - /* Send the DA_ID and Fabric LOGO to cleanup Nameserver entries. */ ndlp = lpfc_findnode_did(vport, Fabric_DID); if (!ndlp) @@ -730,6 +726,10 @@ lpfc_vport_delete(struct fc_vport *fc_vport) skip_logo: + /* Remove FC host to break driver binding. */ + fc_remove_host(shost); + scsi_remove_host(shost); + lpfc_cleanup(vport); /* Remove scsi host now. The nodes are cleaned up. */ -- 2.25.1

2 1

[PATCH openEuler-22.03-LTS-SP1 V1] scsi: lpfc: Move NPIV's transport unregistration to after resource clean up
by Cheng Yu 09 Jun '24

09 Jun '24

From: Justin Tee <justin.tee(a)broadcom.com> mainline inclusion from mainline-v6.9-rc2 commit 4ddf01f2f1504fa08b766e8cfeec558e9f8eef6c category: bugfix bugzilla: https://gitee.com/src-openeuler/kernel/issues/I9UMD0 CVE: CVE-2024-36952 Reference: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?… -------------------------------- There are cases after NPIV deletion where the fabric switch still believes the NPIV is logged into the fabric. This occurs when a vport is unregistered before the Remove All DA_ID CT and LOGO ELS are sent to the fabric. Currently fc_remove_host(), which calls dev_loss_tmo for all D_IDs including the fabric D_ID, removes the last ndlp reference and frees the ndlp rport object. This sometimes causes the race condition where the final DA_ID and LOGO are skipped from being sent to the fabric switch. Fix by moving the fc_remove_host() and scsi_remove_host() calls after DA_ID and LOGO are sent. Signed-off-by: Justin Tee <justin.tee(a)broadcom.com> Link: https://lore.kernel.org/r/20240305200503.57317-3-justintee8345@gmail.com Signed-off-by: Martin K. Petersen <martin.petersen(a)oracle.com> Conflicts: drivers/scsi/lpfc/lpfc_vport.c [There are only context conficts, no code logic conflicts.] Signed-off-by: Cheng Yu <serein.chengyu(a)huawei.com> --- drivers/scsi/lpfc/lpfc_vport.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/drivers/scsi/lpfc/lpfc_vport.c b/drivers/scsi/lpfc/lpfc_vport.c index aa4e451d5dc1..02871d73a502 100644 --- a/drivers/scsi/lpfc/lpfc_vport.c +++ b/drivers/scsi/lpfc/lpfc_vport.c @@ -668,10 +668,6 @@ lpfc_vport_delete(struct fc_vport *fc_vport) ns_ndlp_referenced = true; } - /* Remove FC host and then SCSI host with the vport */ - fc_remove_host(shost); - scsi_remove_host(shost); - ndlp = lpfc_findnode_did(phba->pport, Fabric_DID); /* In case of driver unload, we shall not perform fabric logo as the @@ -774,6 +770,10 @@ lpfc_vport_delete(struct fc_vport *fc_vport) skip_logo: + /* Remove FC host and then SCSI host with the vport */ + fc_remove_host(shost); + scsi_remove_host(shost); + /* * If the NameServer ndlp has been incremented to allow the DA_ID CT * command to be sent, decrement the ndlp now. -- 2.25.1

2 1

[PATCH openEuler-1.0-LTS V1] scsi: lpfc: Move NPIV's transport unregistration to after resource clean up
by Cheng Yu 09 Jun '24

09 Jun '24

From: Justin Tee <justin.tee(a)broadcom.com> mainline inclusion from mainline-v6.9-rc2 commit 4ddf01f2f1504fa08b766e8cfeec558e9f8eef6c category: bugfix bugzilla: https://gitee.com/src-openeuler/kernel/issues/I9UMD0 CVE: CVE-2024-36952 Reference: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?… -------------------------------- There are cases after NPIV deletion where the fabric switch still believes the NPIV is logged into the fabric. This occurs when a vport is unregistered before the Remove All DA_ID CT and LOGO ELS are sent to the fabric. Currently fc_remove_host(), which calls dev_loss_tmo for all D_IDs including the fabric D_ID, removes the last ndlp reference and frees the ndlp rport object. This sometimes causes the race condition where the final DA_ID and LOGO are skipped from being sent to the fabric switch. Fix by moving the fc_remove_host() and scsi_remove_host() calls after DA_ID and LOGO are sent. Signed-off-by: Justin Tee <justin.tee(a)broadcom.com> Link: https://lore.kernel.org/r/20240305200503.57317-3-justintee8345@gmail.com Signed-off-by: Martin K. Petersen <martin.petersen(a)oracle.com> Conflicts: drivers/scsi/lpfc/lpfc_vport.c [There are only context conficts, no code logic conflicts.] Signed-off-by: Cheng Yu <serein.chengyu(a)huawei.com> --- drivers/scsi/lpfc/lpfc_vport.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/drivers/scsi/lpfc/lpfc_vport.c b/drivers/scsi/lpfc/lpfc_vport.c index 1ff0f7de9105..0b60ae657030 100644 --- a/drivers/scsi/lpfc/lpfc_vport.c +++ b/drivers/scsi/lpfc/lpfc_vport.c @@ -690,10 +690,6 @@ lpfc_vport_delete(struct fc_vport *fc_vport) ns_ndlp_referenced = true; } - /* Remove FC host and then SCSI host with the vport */ - fc_remove_host(shost); - scsi_remove_host(shost); - ndlp = lpfc_findnode_did(phba->pport, Fabric_DID); /* In case of driver unload, we shall not perform fabric logo as the @@ -796,6 +792,10 @@ lpfc_vport_delete(struct fc_vport *fc_vport) skip_logo: + /* Remove FC host and then SCSI host with the vport */ + fc_remove_host(shost); + scsi_remove_host(shost); + /* * If the NameServer ndlp has been incremented to allow the DA_ID CT * command to be sent, decrement the ndlp now. -- 2.25.1

2 1

[PATCH OLK-5.10 V1] scsi: lpfc: Move NPIV's transport unregistration to after resource clean up
by Cheng Yu 09 Jun '24

09 Jun '24

From: Justin Tee <justin.tee(a)broadcom.com> mainline inclusion from mainline-v6.9-rc2 commit 4ddf01f2f1504fa08b766e8cfeec558e9f8eef6c category: bugfix bugzilla: https://gitee.com/src-openeuler/kernel/issues/I9UMD0 CVE: CVE-2024-36952 Reference: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?… -------------------------------- There are cases after NPIV deletion where the fabric switch still believes the NPIV is logged into the fabric. This occurs when a vport is unregistered before the Remove All DA_ID CT and LOGO ELS are sent to the fabric. Currently fc_remove_host(), which calls dev_loss_tmo for all D_IDs including the fabric D_ID, removes the last ndlp reference and frees the ndlp rport object. This sometimes causes the race condition where the final DA_ID and LOGO are skipped from being sent to the fabric switch. Fix by moving the fc_remove_host() and scsi_remove_host() calls after DA_ID and LOGO are sent. Signed-off-by: Justin Tee <justin.tee(a)broadcom.com> Link: https://lore.kernel.org/r/20240305200503.57317-3-justintee8345@gmail.com Signed-off-by: Martin K. Petersen <martin.petersen(a)oracle.com> Conflicts: drivers/scsi/lpfc/lpfc_vport.c [There are only context conficts, no code logic conflicts.] Signed-off-by: Cheng Yu <serein.chengyu(a)huawei.com> --- drivers/scsi/lpfc/lpfc_vport.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/drivers/scsi/lpfc/lpfc_vport.c b/drivers/scsi/lpfc/lpfc_vport.c index aa4e451d5dc1..02871d73a502 100644 --- a/drivers/scsi/lpfc/lpfc_vport.c +++ b/drivers/scsi/lpfc/lpfc_vport.c @@ -668,10 +668,6 @@ lpfc_vport_delete(struct fc_vport *fc_vport) ns_ndlp_referenced = true; } - /* Remove FC host and then SCSI host with the vport */ - fc_remove_host(shost); - scsi_remove_host(shost); - ndlp = lpfc_findnode_did(phba->pport, Fabric_DID); /* In case of driver unload, we shall not perform fabric logo as the @@ -774,6 +770,10 @@ lpfc_vport_delete(struct fc_vport *fc_vport) skip_logo: + /* Remove FC host and then SCSI host with the vport */ + fc_remove_host(shost); + scsi_remove_host(shost); + /* * If the NameServer ndlp has been incremented to allow the DA_ID CT * command to be sent, decrement the ndlp now. -- 2.25.1

2 1

[PATCH OLK-5.10] cvm_tsi: enhance security for cvm tsi
by Shengjie Li 09 Jun '24

09 Jun '24

virtcca inclusion category: bugfix bugzilla: https://gitee.com/openeuler/kernel/issues/I9Q85J -------------------------------- enhance security for Confidential cVM tsi: Signed-off-by: Shengjie Li <lishengjie12(a)huawei.com> --- arch/arm64/include/asm/cvm_smc.h | 22 +++---- arch/arm64/include/uapi/asm/cvm_tsi.h | 11 ++-- arch/arm64/kernel/cvm_tsi.c | 89 ++++++++++++++------------- 3 files changed, 64 insertions(+), 58 deletions(-) diff --git a/arch/arm64/include/asm/cvm_smc.h b/arch/arm64/include/asm/cvm_smc.h index c4d56333e3..91688bc169 100644 --- a/arch/arm64/include/asm/cvm_smc.h +++ b/arch/arm64/include/asm/cvm_smc.h @@ -124,30 +124,30 @@ static inline unsigned long tsi_measurement_read(struct cvm_measurement *cvm_mea return res.a0; } -static inline unsigned long tsi_attestation_token_init(struct cvm_attestation_cmd *attest_cmd) +static inline unsigned long tsi_attestation_token_init(unsigned char *challenge) { struct arm_smccc_res res; - unsigned char *challenge; + unsigned char *buf; - challenge = kmalloc(CHALLENGE_SIZE, GFP_KERNEL); - if (!challenge) + buf = kmalloc(CHALLENGE_SIZE, GFP_KERNEL); + if (!buf) return -ENOMEM; - memcpy(challenge, attest_cmd->challenge, CHALLENGE_SIZE); + memcpy(buf, challenge, CHALLENGE_SIZE); - arm_smccc_1_1_smc(SMC_TSI_ATTESTATION_TOKEN_INIT, virt_to_phys(challenge), &res); - kfree(challenge); + arm_smccc_1_1_smc(SMC_TSI_ATTESTATION_TOKEN_INIT, virt_to_phys(buf), &res); + kfree(buf); return res.a0; } -static inline unsigned long tsi_attestation_token_continue(struct cvm_attestation_cmd *attest_cmd) +static inline unsigned long tsi_attestation_token_continue(struct cvm_token_granule *token_granule) { struct arm_smccc_res res; - arm_smccc_1_1_smc(SMC_TSI_ATTESTATION_TOKEN_CONTINUE, virt_to_phys(attest_cmd->granule_ipa), - attest_cmd->offset, attest_cmd->size, &res); + arm_smccc_1_1_smc(SMC_TSI_ATTESTATION_TOKEN_CONTINUE, virt_to_phys(token_granule->ipa), + token_granule->offset, token_granule->size, &res); - attest_cmd->num_wr_bytes = res.a1; + token_granule->num_wr_bytes = res.a1; return res.a0; } diff --git a/arch/arm64/include/uapi/asm/cvm_tsi.h b/arch/arm64/include/uapi/asm/cvm_tsi.h index 43e17a0da0..1ed4386db2 100644 --- a/arch/arm64/include/uapi/asm/cvm_tsi.h +++ b/arch/arm64/include/uapi/asm/cvm_tsi.h @@ -55,10 +55,13 @@ struct cvm_measurement_extend { struct cvm_attestation_cmd { unsigned char challenge[CHALLENGE_SIZE]; /* input: challenge value */ - unsigned long token_size; /* return: challenge value */ - void *granule_head; - void *granule_ipa; /* IPA of the Granule to which the token will be written */ - unsigned long granule_count; + unsigned long token_size; /* return: token size */ +}; + +struct cvm_token_granule { + void *head; + void *ipa; /* IPA of the Granule to which the token will be written */ + unsigned long count; unsigned long offset; /* Offset within Granule to start of buffer in bytes */ unsigned long size; /* Size of buffer in bytes */ unsigned long num_wr_bytes; /* Number of bytes written to buffer */ diff --git a/arch/arm64/kernel/cvm_tsi.c b/arch/arm64/kernel/cvm_tsi.c index b48e5b17f4..252fe5352a 100644 --- a/arch/arm64/kernel/cvm_tsi.c +++ b/arch/arm64/kernel/cvm_tsi.c @@ -4,6 +4,7 @@ #include <linux/slab.h> #include <linux/miscdevice.h> #include <linux/preempt.h> +#include <asm/cvm_guest.h> #include <asm/cvm_smc.h> #include <asm/cvm_tsi.h> @@ -16,20 +17,19 @@ struct attestation_token { static struct attestation_token token; +static DEFINE_SPINLOCK(lock); + static long tmm_tsi_ioctl(struct file *file, unsigned int cmd, unsigned long arg); -static int tmm_tsi_release(struct inode *inode, struct file *file); static ssize_t tmm_token_read(struct file *file, char __user *user_buffer, size_t size, loff_t *offset); static int tmm_get_tsi_version(struct cvm_tsi_version __user *arg); -static int tmm_get_attestation_token(struct cvm_attestation_cmd __user *arg, - struct attestation_token *attest_token); +static int tmm_get_attestation_token(struct cvm_attestation_cmd __user *arg); static int tmm_get_device_cert(struct cca_device_cert __user *arg); static const struct file_operations tmm_tsi_fops = { .owner = THIS_MODULE, .read = tmm_token_read, - .release = tmm_tsi_release, .unlocked_ioctl = tmm_tsi_ioctl }; @@ -44,6 +44,9 @@ static int __init tmm_tsi_init(void) unsigned long ver; int ret; + if (!is_cvm_world()) + return -EIO; + ver = tsi_get_version(); if (ver == SMCCC_RET_NOT_SUPPORTED) { pr_err("tmm_tsi: SMC return not supported!\n"); @@ -56,6 +59,11 @@ static int __init tmm_tsi_init(void) return ret; } + /* Allocate a large memory */ + token.buf = kmalloc(GRANULE_SIZE * MAX_TOKEN_GRANULE_PAGE, GFP_KERNEL); + if (!token.buf) + return -ENOMEM; + pr_warn("tmm_tsi: module loaded (version %lu.%lu).\n", TSI_ABI_VERSION_GET_MAJOR(ver), TSI_ABI_VERSION_GET_MINOR(ver)); @@ -65,8 +73,10 @@ static int __init tmm_tsi_init(void) static void __exit tmm_tsi_exit(void) { - if (token.buf != NULL) + if (token.buf != NULL) { + memset(token.buf, 0, GRANULE_SIZE * MAX_TOKEN_GRANULE_PAGE); kfree(token.buf); + } misc_deregister(&ioctl_dev); pr_warn("tmm_tsi: module unloaded.\n"); } @@ -80,7 +90,7 @@ static long tmm_tsi_ioctl(struct file *file, unsigned int cmd, unsigned long arg ret = tmm_get_tsi_version((struct cvm_tsi_version *)arg); break; case TMM_GET_ATTESTATION_TOKEN: - ret = tmm_get_attestation_token((struct cvm_attestation_cmd *)arg, &token); + ret = tmm_get_attestation_token((struct cvm_attestation_cmd *)arg); break; case TMM_GET_DEVICE_CERT: ret = tmm_get_device_cert((struct cca_device_cert *)arg); @@ -99,28 +109,25 @@ static ssize_t tmm_token_read(struct file *file, char __user *user_buffer, int ret; int to_copy; - if (*offset >= token.size) + spin_lock(&lock); + if (*offset >= token.size) { + spin_unlock(&lock); return 0; + } to_copy = min((int)size, (int)(token.size - *offset)); ret = copy_to_user(user_buffer, token.buf + *offset, to_copy); if (ret) { pr_err("tmm_tsi: copy token to user failed (%d)!\n", ret); + spin_unlock(&lock); return -1; } *offset += to_copy; + spin_unlock(&lock); return to_copy; } -static int tmm_tsi_release(struct inode *inode, struct file *file) -{ - if (token.buf != NULL) { - memset(token.buf, 0, GRANULE_SIZE * MAX_TOKEN_GRANULE_PAGE); - kfree(token.buf); - } - return 0; -} static int tmm_get_tsi_version(struct cvm_tsi_version __user *arg) { @@ -141,65 +148,61 @@ static int tmm_get_tsi_version(struct cvm_tsi_version __user *arg) return 0; } -static int tmm_get_attestation_token(struct cvm_attestation_cmd __user *arg, - struct attestation_token *attest_token) +static int tmm_get_attestation_token(struct cvm_attestation_cmd __user *arg) { unsigned long ret; - struct cvm_attestation_cmd cmd = {0}; + struct cvm_token_granule token_granule = {0}; + unsigned char challenge[CHALLENGE_SIZE]; - ret = copy_from_user(&(cmd.challenge), &(arg->challenge), sizeof(cmd.challenge)); + ret = copy_from_user(challenge, &(arg->challenge), CHALLENGE_SIZE); if (ret) { pr_err("tmm_tsi: copy data from user failed (%lu)!\n", ret); return -EFAULT; } - /* Allocate a large memory */ - attest_token->buf = kmalloc(GRANULE_SIZE * MAX_TOKEN_GRANULE_PAGE, GFP_KERNEL); - if (!attest_token->buf) - return -ENOMEM; - cmd.granule_head = attest_token->buf; - cmd.granule_ipa = cmd.granule_head; + spin_lock(&lock); + token_granule.head = token.buf; + token_granule.ipa = token_granule.head; - /* preempt_disable(); */ - - ret = tsi_attestation_token_init(&cmd); + ret = tsi_attestation_token_init(challenge); if (ret) { pr_err("tmm_tsi: tsi call tsi_attestation_token_init failed (%lu)!\n", ret); + spin_unlock(&lock); return -EIO; } do { /* Retrieve one Granule of data per loop iteration */ - cmd.granule_ipa = cmd.granule_head + - (unsigned long)(cmd.granule_count * GRANULE_SIZE); - cmd.offset = 0; + token_granule.ipa = token_granule.head + + (unsigned long)(token_granule.count * GRANULE_SIZE); + token_granule.offset = 0; do { /* Retrieve sub-Granule chunk of data per loop iteration */ - cmd.size = GRANULE_SIZE - cmd.offset; - ret = tsi_attestation_token_continue(&cmd); - cmd.offset += cmd.num_wr_bytes; - } while (ret == TSI_INCOMPLETE && cmd.offset < GRANULE_SIZE); + token_granule.size = GRANULE_SIZE - token_granule.offset; + ret = tsi_attestation_token_continue(&token_granule); + token_granule.offset += token_granule.num_wr_bytes; + } while (ret == TSI_INCOMPLETE && token_granule.offset < GRANULE_SIZE); - cmd.granule_count += 1; - if (cmd.granule_count >= MAX_TOKEN_GRANULE_PAGE && ret == TSI_INCOMPLETE) { + token_granule.count += 1; + if (token_granule.count >= MAX_TOKEN_GRANULE_PAGE && ret == TSI_INCOMPLETE) { pr_err("tmm_tsi: macro MAX_TOKEN_GRANULE_PAGE (%d) is too small!\n", MAX_TOKEN_GRANULE_PAGE); + spin_unlock(&lock); return -ENOMEM; } } while (ret == TSI_INCOMPLETE); - /* preempt_enable(); */ - /* Send to user space the total size of the token */ - cmd.granule_count = cmd.granule_count - 1; - cmd.token_size = (unsigned long)(GRANULE_SIZE * cmd.granule_count) + cmd.offset; - attest_token->size = cmd.token_size; + token_granule.count = token_granule.count - 1; + token.size = (unsigned long)(GRANULE_SIZE * token_granule.count) + token_granule.offset; - ret = copy_to_user(&(arg->token_size), &(cmd.token_size), sizeof(cmd.token_size)); + ret = copy_to_user(&(arg->token_size), &(token.size), sizeof(token.size)); if (ret) { pr_err("tmm_tsi: copy data to user failed (%lu)!\n", ret); + spin_unlock(&lock); return -EFAULT; } + spin_unlock(&lock); return 0; } -- 2.37.2.windows.2

2 1

[PATCH openEuler-22.03-LTS-SP1] net: core: reject skb_copy(_expand) for fraglist GSO skbs
by Ziyang Xuan 09 Jun '24

09 Jun '24

From: Felix Fietkau <nbd(a)nbd.name> stable inclusion from stable-v5.10.217 commit faa83a7797f06cefed86731ba4baa3b4dfdc06c1 category: bugfix bugzilla: https://gitee.com/src-openeuler/kernel/issues/I9U9P7 CVE: CVE-2024-36929 Reference: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id… -------------------------------- [ Upstream commit d091e579b864fa790dd6a0cd537a22c383126681 ] SKB_GSO_FRAGLIST skbs must not be linearized, otherwise they become invalid. Return NULL if such an skb is passed to skb_copy or skb_copy_expand, in order to prevent a crash on a potential later call to skb_gso_segment. Fixes: 3a1296a38d0c ("net: Support GRO/GSO fraglist chaining.") Signed-off-by: Felix Fietkau <nbd(a)nbd.name> Signed-off-by: David S. Miller <davem(a)davemloft.net> Signed-off-by: Sasha Levin <sashal(a)kernel.org> Signed-off-by: Ziyang Xuan <william.xuanziyang(a)huawei.com> --- net/core/skbuff.c | 27 +++++++++++++++++++-------- 1 file changed, 19 insertions(+), 8 deletions(-) diff --git a/net/core/skbuff.c b/net/core/skbuff.c index 5197b286ef18d..120290a9d2078 100644 --- a/net/core/skbuff.c +++ b/net/core/skbuff.c @@ -1534,11 +1534,17 @@ static inline int skb_alloc_rx_flag(const struct sk_buff *skb) struct sk_buff *skb_copy(const struct sk_buff *skb, gfp_t gfp_mask) { - int headerlen = skb_headroom(skb); - unsigned int size = skb_end_offset(skb) + skb->data_len; - struct sk_buff *n = __alloc_skb(size, gfp_mask, - skb_alloc_rx_flag(skb), NUMA_NO_NODE); + struct sk_buff *n; + unsigned int size; + int headerlen; + + if (WARN_ON_ONCE(skb_shinfo(skb)->gso_type & SKB_GSO_FRAGLIST)) + return NULL; + headerlen = skb_headroom(skb); + size = skb_end_offset(skb) + skb->data_len; + n = __alloc_skb(size, gfp_mask, + skb_alloc_rx_flag(skb), NUMA_NO_NODE); if (!n) return NULL; @@ -1799,12 +1805,17 @@ struct sk_buff *skb_copy_expand(const struct sk_buff *skb, /* * Allocate the copy buffer */ - struct sk_buff *n = __alloc_skb(newheadroom + skb->len + newtailroom, - gfp_mask, skb_alloc_rx_flag(skb), - NUMA_NO_NODE); - int oldheadroom = skb_headroom(skb); int head_copy_len, head_copy_off; + struct sk_buff *n; + int oldheadroom; + + if (WARN_ON_ONCE(skb_shinfo(skb)->gso_type & SKB_GSO_FRAGLIST)) + return NULL; + oldheadroom = skb_headroom(skb); + n = __alloc_skb(newheadroom + skb->len + newtailroom, + gfp_mask, skb_alloc_rx_flag(skb), + NUMA_NO_NODE); if (!n) return NULL; -- 2.25.1

2 1

[PATCH OLK-5.10] net: core: reject skb_copy(_expand) for fraglist GSO skbs
by Ziyang Xuan 09 Jun '24

09 Jun '24

From: Felix Fietkau <nbd(a)nbd.name> stable inclusion from stable-v5.10.217 commit faa83a7797f06cefed86731ba4baa3b4dfdc06c1 category: bugfix bugzilla: https://gitee.com/src-openeuler/kernel/issues/I9U9P7 CVE: CVE-2024-36929 Reference: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id… -------------------------------- [ Upstream commit d091e579b864fa790dd6a0cd537a22c383126681 ] SKB_GSO_FRAGLIST skbs must not be linearized, otherwise they become invalid. Return NULL if such an skb is passed to skb_copy or skb_copy_expand, in order to prevent a crash on a potential later call to skb_gso_segment. Fixes: 3a1296a38d0c ("net: Support GRO/GSO fraglist chaining.") Signed-off-by: Felix Fietkau <nbd(a)nbd.name> Signed-off-by: David S. Miller <davem(a)davemloft.net> Signed-off-by: Sasha Levin <sashal(a)kernel.org> Signed-off-by: Ziyang Xuan <william.xuanziyang(a)huawei.com> --- net/core/skbuff.c | 27 +++++++++++++++++++-------- 1 file changed, 19 insertions(+), 8 deletions(-) diff --git a/net/core/skbuff.c b/net/core/skbuff.c index 779a860e00731..de0229b8a920c 100644 --- a/net/core/skbuff.c +++ b/net/core/skbuff.c @@ -1540,11 +1540,17 @@ static inline int skb_alloc_rx_flag(const struct sk_buff *skb) struct sk_buff *skb_copy(const struct sk_buff *skb, gfp_t gfp_mask) { - int headerlen = skb_headroom(skb); - unsigned int size = skb_end_offset(skb) + skb->data_len; - struct sk_buff *n = __alloc_skb(size, gfp_mask, - skb_alloc_rx_flag(skb), NUMA_NO_NODE); + struct sk_buff *n; + unsigned int size; + int headerlen; + + if (WARN_ON_ONCE(skb_shinfo(skb)->gso_type & SKB_GSO_FRAGLIST)) + return NULL; + headerlen = skb_headroom(skb); + size = skb_end_offset(skb) + skb->data_len; + n = __alloc_skb(size, gfp_mask, + skb_alloc_rx_flag(skb), NUMA_NO_NODE); if (!n) return NULL; @@ -1856,12 +1862,17 @@ struct sk_buff *skb_copy_expand(const struct sk_buff *skb, /* * Allocate the copy buffer */ - struct sk_buff *n = __alloc_skb(newheadroom + skb->len + newtailroom, - gfp_mask, skb_alloc_rx_flag(skb), - NUMA_NO_NODE); - int oldheadroom = skb_headroom(skb); int head_copy_len, head_copy_off; + struct sk_buff *n; + int oldheadroom; + + if (WARN_ON_ONCE(skb_shinfo(skb)->gso_type & SKB_GSO_FRAGLIST)) + return NULL; + oldheadroom = skb_headroom(skb); + n = __alloc_skb(newheadroom + skb->len + newtailroom, + gfp_mask, skb_alloc_rx_flag(skb), + NUMA_NO_NODE); if (!n) return NULL; -- 2.25.1

2 1

[PATCH OLK-6.6] net: core: reject skb_copy(_expand) for fraglist GSO skbs
by Ziyang Xuan 09 Jun '24

09 Jun '24

From: Felix Fietkau <nbd(a)nbd.name> stable inclusion from stable-v6.6.31 commit cfe34d86ef9765c388f145039006bb79b6c81ac6 category: bugfix bugzilla: https://gitee.com/src-openeuler/kernel/issues/I9U9P7 CVE: CVE-2024-36929 Reference: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id… -------------------------------- [ Upstream commit d091e579b864fa790dd6a0cd537a22c383126681 ] SKB_GSO_FRAGLIST skbs must not be linearized, otherwise they become invalid. Return NULL if such an skb is passed to skb_copy or skb_copy_expand, in order to prevent a crash on a potential later call to skb_gso_segment. Fixes: 3a1296a38d0c ("net: Support GRO/GSO fraglist chaining.") Signed-off-by: Felix Fietkau <nbd(a)nbd.name> Signed-off-by: David S. Miller <davem(a)davemloft.net> Signed-off-by: Sasha Levin <sashal(a)kernel.org> Signed-off-by: Ziyang Xuan <william.xuanziyang(a)huawei.com> --- net/core/skbuff.c | 27 +++++++++++++++++++-------- 1 file changed, 19 insertions(+), 8 deletions(-) diff --git a/net/core/skbuff.c b/net/core/skbuff.c index 60876262b3fb3..f0a9ef1aeaa29 100644 --- a/net/core/skbuff.c +++ b/net/core/skbuff.c @@ -1971,11 +1971,17 @@ static inline int skb_alloc_rx_flag(const struct sk_buff *skb) struct sk_buff *skb_copy(const struct sk_buff *skb, gfp_t gfp_mask) { - int headerlen = skb_headroom(skb); - unsigned int size = skb_end_offset(skb) + skb->data_len; - struct sk_buff *n = __alloc_skb(size, gfp_mask, - skb_alloc_rx_flag(skb), NUMA_NO_NODE); + struct sk_buff *n; + unsigned int size; + int headerlen; + + if (WARN_ON_ONCE(skb_shinfo(skb)->gso_type & SKB_GSO_FRAGLIST)) + return NULL; + headerlen = skb_headroom(skb); + size = skb_end_offset(skb) + skb->data_len; + n = __alloc_skb(size, gfp_mask, + skb_alloc_rx_flag(skb), NUMA_NO_NODE); if (!n) return NULL; @@ -2303,12 +2309,17 @@ struct sk_buff *skb_copy_expand(const struct sk_buff *skb, /* * Allocate the copy buffer */ - struct sk_buff *n = __alloc_skb(newheadroom + skb->len + newtailroom, - gfp_mask, skb_alloc_rx_flag(skb), - NUMA_NO_NODE); - int oldheadroom = skb_headroom(skb); int head_copy_len, head_copy_off; + struct sk_buff *n; + int oldheadroom; + + if (WARN_ON_ONCE(skb_shinfo(skb)->gso_type & SKB_GSO_FRAGLIST)) + return NULL; + oldheadroom = skb_headroom(skb); + n = __alloc_skb(newheadroom + skb->len + newtailroom, + gfp_mask, skb_alloc_rx_flag(skb), + NUMA_NO_NODE); if (!n) return NULL; -- 2.25.1

2 1

[openeuler:OLK-5.10] BUILD SUCCESS 9e1339f3705f9326a4594db2783ca5d7522d1a9c
by kernel test robot 09 Jun '24

09 Jun '24

tree/branch: https://gitee.com/openeuler/kernel.git OLK-5.10 branch HEAD: 9e1339f3705f9326a4594db2783ca5d7522d1a9c !8684 tcp: properly terminate timers for kernel sockets Warning ids grouped by kconfigs: clang_recent_errors `-- x86_64-allnoconfig |-- drivers-arm-spe-spe.c:linux-perf-arm_pmu.h-is-included-more-than-once. |-- drivers-net-ethernet-yunsilicon-xsc-net-main.c:common-qp.h-is-included-more-than-once. |-- drivers-ub-urma-ubcore-ubcore_cdev_file.c:linux-version.h-not-needed. |-- drivers-ub-urma-ubcore-ubcore_device.c:linux-version.h-not-needed. |-- drivers-ub-urma-ubcore-ubcore_genl.c:linux-version.h-not-needed. |-- drivers-ub-urma-ubcore-ubcore_genl_admin.c:linux-version.h-not-needed. |-- drivers-ub-urma-ubcore-ubcore_uvs_cmd.c:ubcore_device.h-is-included-more-than-once. `-- drivers-ub-urma-uburma-uburma_mmap.c:linux-version.h-not-needed. elapsed time: 1237m configs tested: 35 configs skipped: 131 The following configs have been built successfully. More configs may be tested in the coming days. tested configs: arm64 allmodconfig clang arm64 allnoconfig gcc arm64 defconfig gcc arm64 randconfig-001-20240609 gcc arm64 randconfig-002-20240609 clang arm64 randconfig-003-20240609 gcc arm64 randconfig-004-20240609 gcc x86_64 allnoconfig clang x86_64 allyesconfig clang x86_64 buildonly-randconfig-001-20240609 clang x86_64 buildonly-randconfig-002-20240609 clang x86_64 buildonly-randconfig-003-20240609 clang x86_64 buildonly-randconfig-004-20240609 gcc x86_64 buildonly-randconfig-005-20240609 clang x86_64 buildonly-randconfig-006-20240609 gcc x86_64 defconfig gcc x86_64 randconfig-001-20240609 gcc x86_64 randconfig-002-20240609 clang x86_64 randconfig-003-20240609 gcc x86_64 randconfig-004-20240609 gcc x86_64 randconfig-005-20240609 clang x86_64 randconfig-006-20240609 clang x86_64 randconfig-011-20240609 gcc x86_64 randconfig-012-20240609 gcc x86_64 randconfig-013-20240609 clang x86_64 randconfig-014-20240609 gcc x86_64 randconfig-015-20240609 clang x86_64 randconfig-016-20240609 gcc x86_64 randconfig-071-20240609 clang x86_64 randconfig-072-20240609 clang x86_64 randconfig-073-20240609 gcc x86_64 randconfig-074-20240609 clang x86_64 randconfig-075-20240609 gcc x86_64 randconfig-076-20240609 gcc x86_64 rhel-8.3-rust clang -- 0-DAY CI Kernel Test Service https://github.com/intel/lkp-tests/wiki

1 0