- Linux-stable-mirror - lists.linaro.org

[PATCH 5.15.y] page_pool: Fix use-after-free in page_pool_recycle_in_ring

by ruohanlan＠aliyun.com

From: Dong Chenchen <dongchenchen2(a)huawei.com> [ Upstream commit 271683bb2cf32e5126c592b5d5e6a756fa374fd9 ] syzbot reported a uaf in page_pool_recycle_in_ring: BUG: KASAN: slab-use-after-free in lock_release+0x151/0xa30 kernel/locking/lockdep.c:5862 Read of size 8 at addr ffff8880286045a0 by task syz.0.284/6943 CPU: 0 UID: 0 PID: 6943 Comm: syz.0.284 Not tainted 6.13.0-rc3-syzkaller-gdfa94ce54f41 #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 09/13/2024 Call Trace: <TASK> __dump_stack lib/dump_stack.c:94 [inline] dump_stack_lvl+0x241/0x360 lib/dump_stack.c:120 print_address_description mm/kasan/report.c:378 [inline] print_report+0x169/0x550 mm/kasan/report.c:489 kasan_report+0x143/0x180 mm/kasan/report.c:602 lock_release+0x151/0xa30 kernel/locking/lockdep.c:5862 __raw_spin_unlock_bh include/linux/spinlock_api_smp.h:165 [inline] _raw_spin_unlock_bh+0x1b/0x40 kernel/locking/spinlock.c:210 spin_unlock_bh include/linux/spinlock.h:396 [inline] ptr_ring_produce_bh include/linux/ptr_ring.h:164 [inline] page_pool_recycle_in_ring net/core/page_pool.c:707 [inline] page_pool_put_unrefed_netmem+0x748/0xb00 net/core/page_pool.c:826 page_pool_put_netmem include/net/page_pool/helpers.h:323 [inline] page_pool_put_full_netmem include/net/page_pool/helpers.h:353 [inline] napi_pp_put_page+0x149/0x2b0 net/core/skbuff.c:1036 skb_pp_recycle net/core/skbuff.c:1047 [inline] skb_free_head net/core/skbuff.c:1094 [inline] skb_release_data+0x6c4/0x8a0 net/core/skbuff.c:1125 skb_release_all net/core/skbuff.c:1190 [inline] __kfree_skb net/core/skbuff.c:1204 [inline] sk_skb_reason_drop+0x1c9/0x380 net/core/skbuff.c:1242 kfree_skb_reason include/linux/skbuff.h:1263 [inline] __skb_queue_purge_reason include/linux/skbuff.h:3343 [inline] root cause is: page_pool_recycle_in_ring ptr_ring_produce spin_lock(&r->producer_lock); WRITE_ONCE(r->queue[r->producer++], ptr) //recycle last page to pool page_pool_release page_pool_scrub page_pool_empty_ring ptr_ring_consume page_pool_return_page //release all page __page_pool_destroy free_percpu(pool->recycle_stats); free(pool) //free spin_unlock(&r->producer_lock); //pool->ring uaf read recycle_stat_inc(pool, ring); page_pool can be free while page pool recycle the last page in ring. Add producer-lock barrier to page_pool_release to prevent the page pool from being free before all pages have been recycled. recycle_stat_inc() is empty when CONFIG_PAGE_POOL_STATS is not enabled, which will trigger Wempty-body build warning. Add definition for pool stat macro to fix warning. Suggested-by: Jakub Kicinski <kuba(a)kernel.org> Link: https://lore.kernel.org/netdev/20250513083123.3514193-1-dongchenchen2@huawe… Fixes: ff7d6b27f894 ("page_pool: refurbish version of page_pool code") Reported-by: syzbot+204a4382fcb3311f3858(a)syzkaller.appspotmail.com Closes: https://syzkaller.appspot.com/bug?extid=204a4382fcb3311f3858 Signed-off-by: Dong Chenchen <dongchenchen2(a)huawei.com> Reviewed-by: Toke Høiland-Jørgensen <toke(a)redhat.com> Reviewed-by: Mina Almasry <almasrymina(a)google.com> Link: https://patch.msgid.link/20250527114152.3119109-1-dongchenchen2@huawei.com Signed-off-by: Jakub Kicinski <kuba(a)kernel.org> [ Linux 5.15.y does not support page pool statistics (CONFIG_PAGE_POOL_STATS), so remove the related source code changes from the patch. ] Signed-off-by: Ruohan Lan <ruohanlan(a)aliyun.com> --- net/core/page_pool.c | 16 ++++++++++------ 1 file changed, 10 insertions(+), 6 deletions(-) diff --git a/net/core/page_pool.c b/net/core/page_pool.c index 4ec7e6229f69..b3047197a5e8 100644 --- a/net/core/page_pool.c +++ b/net/core/page_pool.c @@ -414,14 +414,14 @@ static void page_pool_return_page(struct page_pool *pool, struct page *page) static bool page_pool_recycle_in_ring(struct page_pool *pool, struct page *page) { - int ret; + bool in_softirq, ret; + /* BH protection not needed if current is softirq */ - if (in_softirq()) - ret = ptr_ring_produce(&pool->ring, page); - else - ret = ptr_ring_produce_bh(&pool->ring, page); + in_softirq = page_pool_producer_lock(pool); + ret = !__ptr_ring_produce(&pool->ring, page); + page_pool_producer_unlock(pool, in_softirq); - return (ret == 0) ? true : false; + return ret; } /* Only allow direct recycling in special circumstances, into the @@ -684,10 +684,14 @@ static void page_pool_scrub(struct page_pool *pool) static int page_pool_release(struct page_pool *pool) { + bool in_softirq; int inflight; page_pool_scrub(pool); inflight = page_pool_inflight(pool); + /* Acquire producer lock to make sure producers have exited. */ + in_softirq = page_pool_producer_lock(pool); + page_pool_producer_unlock(pool, in_softirq); if (!inflight) page_pool_free(pool); -- 2.43.0

4 days, 20 hours

1
0
0 0

[PATCH] drm/msm/dpu: Add missing NULL pointer check for pingpong interface

by Nikolay Kuratov

It is checked almost always in dpu_encoder_phys_wb_setup_ctl(), but in a single place the check is missing. Also use convenient locals instead of phys_enc->* where available. Cc: stable(a)vger.kernel.org Fixes: d7d0e73f7de33 ("drm/msm/dpu: introduce the dpu_encoder_phys_* for writeback") Signed-off-by: Nikolay Kuratov <kniv(a)yandex-team.ru> --- drivers/gpu/drm/msm/disp/dpu1/dpu_encoder_phys_wb.c | 10 ++++------ 1 file changed, 4 insertions(+), 6 deletions(-) diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder_phys_wb.c b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder_phys_wb.c index 46f348972a97..6d28f2281c76 100644 --- a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder_phys_wb.c +++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder_phys_wb.c @@ -247,14 +247,12 @@ static void dpu_encoder_phys_wb_setup_ctl(struct dpu_encoder_phys *phys_enc) if (hw_cdm) intf_cfg.cdm = hw_cdm->idx; - if (phys_enc->hw_pp->merge_3d && phys_enc->hw_pp->merge_3d->ops.setup_3d_mode) - phys_enc->hw_pp->merge_3d->ops.setup_3d_mode(phys_enc->hw_pp->merge_3d, - mode_3d); + if (hw_pp && hw_pp->merge_3d && hw_pp->merge_3d->ops.setup_3d_mode) + hw_pp->merge_3d->ops.setup_3d_mode(hw_pp->merge_3d, mode_3d); /* setup which pp blk will connect to this wb */ - if (hw_pp && phys_enc->hw_wb->ops.bind_pingpong_blk) - phys_enc->hw_wb->ops.bind_pingpong_blk(phys_enc->hw_wb, - phys_enc->hw_pp->idx); + if (hw_pp && hw_wb->ops.bind_pingpong_blk) + hw_wb->ops.bind_pingpong_blk(hw_wb, hw_pp->idx); phys_enc->hw_ctl->ops.setup_intf_cfg(phys_enc->hw_ctl, &intf_cfg); } else if (phys_enc->hw_ctl && phys_enc->hw_ctl->ops.setup_intf_cfg) { -- 2.34.1

4 days, 20 hours

2
1
0 0

[PATCH net v3 1/1] atm: mpoa: Fix UAF on qos_head list in procfs

by Minseong Kim

The global QoS list 'qos_head' in net/atm/mpc.c is accessed from the /proc/net/atm/mpc procfs interface without proper synchronization. The read-side seq_file show path (mpc_show() -> atm_mpoa_disp_qos()) walks qos_head without any lock, while the write-side path (proc_mpc_write() -> parse_qos() -> atm_mpoa_delete_qos()) can unlink and kfree() entries immediately. Concurrent read/write therefore leads to a use-after-free. Fix this by serializing all qos_head list operations with a mutex. A mutex is used because seq_printf() may sleep. To avoid returning an unprotected pointer (and the resulting lifetime race), replace atm_mpoa_search_qos() with atm_mpoa_get_qos(), which copies the QoS under lock. Also add atm_mpoa_delete_qos_by_ip() so search+unlink+free is atomic under the same lock, preventing concurrent double-free/UAF scenarios. KASAN report: [ 15.911538] BUG: KASAN: slab-use-after-free in atm_mpoa_disp_qos+0x202/0x380 [ 15.911988] Read of size 8 at addr ffff888007517380 by task mpoa_uaf_poc/89 [ 15.912123] [ 15.912717] CPU: 0 UID: 0 PID: 89 Comm: mpoa_uaf_poc Not tainted 6.17.8 #1 PREEMPT(voluntary) [ 15.912950] Hardware name: QEMU Ubuntu 25.04 PC (i440FX + PIIX, 1996), BIOS 1.16.3-debian-1.16.3-2 04/01/2014 [ 15.913110] Call Trace: [ 15.913167] <TASK> [ 15.913247] dump_stack_lvl+0x4e/0x70 [ 15.913343] ? atm_mpoa_disp_qos+0x202/0x380 [ 15.913364] print_report+0x174/0x4f6 [ 15.913386] ? __pfx__raw_spin_lock_irqsave+0x10/0x10 [ 15.913412] ? atm_mpoa_disp_qos+0x202/0x380 [ 15.913430] kasan_report+0xce/0x100 [ 15.913453] ? atm_mpoa_disp_qos+0x202/0x380 [ 15.913475] atm_mpoa_disp_qos+0x202/0x380 [ 15.913496] mpc_show+0x575/0x700 [ 15.913515] ? kasan_save_track+0x14/0x30 [ 15.913532] ? __pfx_mpc_show+0x10/0x10 [ 15.913550] ? __pfx_mutex_lock+0x10/0x10 [ 15.913565] ? seq_read_iter+0x697/0x1110 [ 15.913584] seq_read_iter+0x2bc/0x1110 [ 15.913607] seq_read+0x267/0x3d0 [ 15.913623] ? __pfx_seq_read+0x10/0x10 [ 15.913650] proc_reg_read+0x1ab/0x270 [ 15.913669] vfs_read+0x175/0xa10 [ 15.913687] ? do_sys_openat2+0x103/0x170 [ 15.913701] ? kmem_cache_free+0xc4/0x360 [ 15.913718] ? getname_flags.part.0+0xf3/0x470 [ 15.913734] ? __pfx_vfs_read+0x10/0x10 [ 15.913751] ? mutex_lock+0x81/0xe0 [ 15.913766] ? __pfx_mutex_lock+0x10/0x10 [ 15.913782] ? __rseq_handle_notify_resume+0x4c4/0xac0 [ 15.913802] ? fdget_pos+0x24d/0x4b0 [ 15.913829] ksys_read+0xf7/0x1c0 [ 15.913850] ? __pfx_ksys_read+0x10/0x10 [ 15.913877] do_syscall_64+0xa4/0x290 [ 15.913917] entry_SYSCALL_64_after_hwframe+0x77/0x7f [ 15.913981] RIP: 0033:0x45c982 [ 15.914171] Code: 08 0f 85 71 ea ff ff 49 89 fb 48 89 f0 48 89 d7 48 89 ce 4c 89 c2 4d 89 ca 4c 8b 44 24 08 4c 8b 4c 24 10 4c 89 5c 24 08 0f 05 <c3> 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 00 f3 0f 1e fa 55 48 89 e5 [ 15.914239] RSP: 002b:00007f4b2b2cb0d8 EFLAGS: 00000246 ORIG_RAX: 0000000000000000 [ 15.914315] RAX: ffffffffffffffda RBX: 00007f4b2b2cc6c0 RCX: 000000000045c982 [ 15.914352] RDX: 0000000000000fff RSI: 00007f4b2b2cb220 RDI: 0000000000000014 [ 15.914382] RBP: 00007f4b2b2cb100 R08: 0000000000000000 R09: 0000000000000000 [ 15.914413] R10: 0000000000000000 R11: 0000000000000246 R12: 000000000000000a [ 15.914445] R13: ffffffffffffffd0 R14: 0000000000000000 R15: 00007ffd386450c0 [ 15.914483] </TASK> [ 15.914533] [ 15.916812] Allocated by task 78: [ 15.916975] kasan_save_stack+0x30/0x50 [ 15.917047] kasan_save_track+0x14/0x30 [ 15.917105] __kasan_kmalloc+0x8f/0xa0 [ 15.917162] atm_mpoa_add_qos+0x1bb/0x3c0 [ 15.917220] parse_qos.cold+0x73/0x7d [ 15.917276] proc_mpc_write+0xf4/0x150 [ 15.917331] proc_reg_write+0x1ab/0x270 [ 15.917379] vfs_write+0x1ce/0xd30 [ 15.917431] ksys_write+0xf7/0x1c0 [ 15.917476] do_syscall_64+0xa4/0x290 [ 15.917523] entry_SYSCALL_64_after_hwframe+0x77/0x7f [ 15.917586] [ 15.917628] Freed by task 78: [ 15.917665] kasan_save_stack+0x30/0x50 [ 15.917714] kasan_save_track+0x14/0x30 [ 15.917762] kasan_save_free_info+0x3b/0x70 [ 15.917811] __kasan_slab_free+0x3e/0x50 [ 15.918058] kfree+0x121/0x340 [ 15.918135] atm_mpoa_delete_qos+0xad/0xd0 [ 15.918196] parse_qos+0x1e5/0x1f0 [ 15.918339] proc_mpc_write+0xf4/0x150 [ 15.918438] proc_reg_write+0x1ab/0x270 [ 15.918494] vfs_write+0x1ce/0xd30 [ 15.918538] ksys_write+0xf7/0x1c0 [ 15.918589] do_syscall_64+0xa4/0x290 [ 15.918634] entry_SYSCALL_64_after_hwframe+0x77/0x7f [ 15.918694] [ 15.918751] The buggy address belongs to the object at ffff888007517380 [ 15.918751] which belongs to the cache kmalloc-96 of size 96 [ 15.918911] The buggy address is located 0 bytes inside of [ 15.918911] freed 96-byte region [ffff888007517380, ffff8880075173e0) [ 15.919027] [ 15.919101] The buggy address belongs to the physical page: [ 15.919416] page: refcount:0 mapcount:0 mapping:0000000000000000 index:0xffff888007517300 pfn:0x7517 [ 15.919679] anon flags: 0x100000000000000(node=0|zone=1) [ 15.920064] page_type: f5(slab) [ 15.920361] raw: 0100000000000000 ffff888006c41280 0000000000000000 dead000000000001 [ 15.920460] raw: ffff888007517300 0000000080200007 00000000f5000000 0000000000000000 [ 15.920577] page dumped because: kasan: bad access detected [ 15.920634] [ 15.920663] Memory state around the buggy address: [ 15.920860] ffff888007517280: fa fb fb fb fb fb fb fb fb fb fb fb fc fc fc fc [ 15.920944] ffff888007517300: fa fb fb fb fb fb fb fb fb fb fb fb fc fc fc fc [ 15.921017] >ffff888007517380: fa fb fb fb fb fb fb fb fb fb fb fb fc fc fc fc [ 15.921088] ^ [ 15.921163] ffff888007517400: fa fb fb fb fb fb fb fb fb fb fb fb fc fc fc fc [ 15.921222] ffff888007517480: fa fb fb fb fb fb fb fb fb fb fb fb fc fc fc fc Reported-by: Minseong Kim <ii4gsp(a)gmail.com> Closes: https://lore.kernel.org/netdev/CAKrymDR1X3XTX_1ZW3XXXnuYH+kzsnv7Av5uivzR1st… Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Cc: stable(a)vger.kernel.org Signed-off-by: Minseong Kim <ii4gsp(a)gmail.com> --- net/atm/mpc.c | 184 ++++++++++++++++++++++++++++++++---------- net/atm/mpc.h | 3 +- net/atm/mpoa_caches.c | 17 ++-- net/atm/mpoa_proc.c | 2 +- 4 files changed, 149 insertions(+), 57 deletions(-) diff --git a/net/atm/mpc.c b/net/atm/mpc.c index f6b447bba329..650081dd82d1 100644 --- a/net/atm/mpc.c +++ b/net/atm/mpc.c @@ -9,6 +9,7 @@ #include <linux/bitops.h> #include <linux/capability.h> #include <linux/seq_file.h> +#include <linux/mutex.h> /* We are an ethernet device */ #include <linux/if_ether.h> @@ -122,6 +123,7 @@ static struct notifier_block mpoa_notifier = { struct mpoa_client *mpcs = NULL; /* FIXME */ static struct atm_mpoa_qos *qos_head = NULL; +static DEFINE_MUTEX(qos_mutex); /* Protect qos_head list */ static DEFINE_TIMER(mpc_timer, mpc_cache_check); @@ -172,67 +174,63 @@ static struct mpoa_client *find_mpc_by_lec(struct net_device *dev) */ /* - * Overwrites the old entry or makes a new one. + * Search for a QoS entry. Caller must hold qos_mutex. + * Returns pointer to entry if found, NULL otherwise. */ -struct atm_mpoa_qos *atm_mpoa_add_qos(__be32 dst_ip, struct atm_qos *qos) +static struct atm_mpoa_qos *__atm_mpoa_search_qos(__be32 dst_ip) { - struct atm_mpoa_qos *entry; - - entry = atm_mpoa_search_qos(dst_ip); - if (entry != NULL) { - entry->qos = *qos; - return entry; - } + struct atm_mpoa_qos *qos = qos_head; - entry = kmalloc(sizeof(struct atm_mpoa_qos), GFP_KERNEL); - if (entry == NULL) { - pr_info("mpoa: out of memory\n"); - return entry; + while (qos) { + if (qos->ipaddr == dst_ip) + return qos; + qos = qos->next; } - - entry->ipaddr = dst_ip; - entry->qos = *qos; - - entry->next = qos_head; - qos_head = entry; - - return entry; + return NULL; } -struct atm_mpoa_qos *atm_mpoa_search_qos(__be32 dst_ip) +/* + * Get a QoS entry by copying its value. + * Caller gets a COPY of qos under lock. + * Returns true if found and copied, false if not found. + * This avoids lifetime issues with the returned pointer. + */ +bool atm_mpoa_get_qos(__be32 dst_ip, struct atm_qos *out) { - struct atm_mpoa_qos *qos; + struct atm_mpoa_qos *q; - qos = qos_head; - while (qos) { - if (qos->ipaddr == dst_ip) - break; - qos = qos->next; - } + if (!out) + return false; - return qos; + mutex_lock(&qos_mutex); + q = __atm_mpoa_search_qos(dst_ip); + if (q) + *out = q->qos; + mutex_unlock(&qos_mutex); + + return q; } /* - * Returns 0 for failure + * Delete a QoS entry from the list. Caller must hold qos_mutex. + * Returns 1 if found and deleted, 0 if not found. */ -int atm_mpoa_delete_qos(struct atm_mpoa_qos *entry) +static int __atm_mpoa_delete_qos_locked(struct atm_mpoa_qos *entry) { struct atm_mpoa_qos *curr; - if (entry == NULL) + if (!entry) return 0; + if (entry == qos_head) { - qos_head = qos_head->next; - kfree(entry); + qos_head = entry->next; return 1; } curr = qos_head; - while (curr != NULL) { + while (curr) { if (curr->next == entry) { curr->next = entry->next; - kfree(entry); return 1; } curr = curr->next; @@ -241,15 +239,107 @@ int atm_mpoa_delete_qos(struct atm_mpoa_qos *entry) return 0; } +/* + * Delete a QoS entry by IP address. + * This is atomic: search+unlink+free happens entirely under lock, + * avoiding the double-free issue with atm_mpoa_delete_qos(atm_mpoa_search_qos(ipaddr)). + */ +int atm_mpoa_delete_qos_by_ip(__be32 dst_ip) +{ + struct atm_mpoa_qos *entry; + int ret = 0; + + mutex_lock(&qos_mutex); + entry = __atm_mpoa_search_qos(dst_ip); + if (entry) { + ret = __atm_mpoa_delete_qos_locked(entry); + if (ret) + kfree(entry); + } + mutex_unlock(&qos_mutex); + + return ret; +} + +/* + * Overwrites the old entry or makes a new one. + */ +struct atm_mpoa_qos *atm_mpoa_add_qos(__be32 dst_ip, struct atm_qos *qos) +{ + struct atm_mpoa_qos *entry; + struct atm_mpoa_qos *new; + + /* Fast path: update existing entry */ + mutex_lock(&qos_mutex); + entry = __atm_mpoa_search_qos(dst_ip); + if (entry) { + entry->qos = *qos; + mutex_unlock(&qos_mutex); + return entry; + } + mutex_unlock(&qos_mutex); + + /* Allocate outside lock */ + new = kmalloc(sizeof(*new), GFP_KERNEL); + if (!new) + return NULL; + + new->ipaddr = dst_ip; + new->qos = *qos; + + /* Re-check under lock to avoid duplicates */ + mutex_lock(&qos_mutex); + entry = __atm_mpoa_search_qos(dst_ip); + if (entry) { + entry->qos = *qos; + mutex_unlock(&qos_mutex); + kfree(new); + return entry; + } + + new->next = qos_head; + qos_head = new; + mutex_unlock(&qos_mutex); + + return new; +} + +/* + * Delete a QoS entry by pointer. + * WARNING: This function is unsafe if called with an entry pointer + * obtained outside of qos_mutex protection. The entry may have been + * freed by another thread between the time the pointer was obtained + * and when this function is called. + * Use atm_mpoa_delete_qos_by_ip() instead for safe deletion by IP address. + * Returns 0 for failure, 1 for success + */ +int atm_mpoa_delete_qos(struct atm_mpoa_qos *entry) +{ + int ret; + + if (!entry) + return 0; + + mutex_lock(&qos_mutex); + ret = __atm_mpoa_delete_qos_locked(entry); + if (ret) + kfree(entry); + mutex_unlock(&qos_mutex); + + return ret; +} + /* this is buggered - we need locking for qos_head */ void atm_mpoa_disp_qos(struct seq_file *m) { struct atm_mpoa_qos *qos; - qos = qos_head; seq_printf(m, "QoS entries for shortcuts:\n"); - seq_printf(m, "IP address\n TX:max_pcr pcr min_pcr max_cdv max_sdu\n RX:max_pcr pcr min_pcr max_cdv max_sdu\n"); + seq_printf(m, "IP address\n TX:max_pcr pcr min_pcr max_cdv max_sdu\n" + " RX:max_pcr pcr min_pcr max_cdv max_sdu\n"); + mutex_lock(&qos_mutex); + qos = qos_head; while (qos != NULL) { seq_printf(m, "%pI4\n %-7d %-7d %-7d %-7d %-7d\n %-7d %-7d %-7d %-7d %-7d\n", &qos->ipaddr, @@ -265,6 +355,7 @@ void atm_mpoa_disp_qos(struct seq_file *m) qos->qos.rxtp.max_sdu); qos = qos->next; } + mutex_unlock(&qos_mutex); } static struct net_device *find_lec_by_itfnum(int itf) @@ -1118,13 +1209,16 @@ static void check_qos_and_open_shortcut(struct k_message *msg, in_cache_entry *entry) { __be32 dst_ip = msg->content.in_info.in_dst_ip; - struct atm_mpoa_qos *qos = atm_mpoa_search_qos(dst_ip); + struct atm_qos tmp_qos; + bool has_qos; eg_cache_entry *eg_entry = client->eg_ops->get_by_src_ip(dst_ip, client); + has_qos = atm_mpoa_get_qos(dst_ip, &tmp_qos); + if (eg_entry && eg_entry->shortcut) { if (eg_entry->shortcut->qos.txtp.traffic_class & msg->qos.txtp.traffic_class & - (qos ? qos->qos.txtp.traffic_class : ATM_UBR | ATM_CBR)) { + (has_qos ? tmp_qos.txtp.traffic_class : ATM_UBR | ATM_CBR)) { if (eg_entry->shortcut->qos.txtp.traffic_class == ATM_UBR) entry->shortcut = eg_entry->shortcut; else if (eg_entry->shortcut->qos.txtp.max_pcr > 0) @@ -1142,9 +1236,9 @@ static void check_qos_and_open_shortcut(struct k_message *msg, /* No luck in the egress cache we must open an ingress SVC */ msg->type = OPEN_INGRESS_SVC; - if (qos && - (qos->qos.txtp.traffic_class == msg->qos.txtp.traffic_class)) { - msg->qos = qos->qos; + if (has_qos && + tmp_qos.txtp.traffic_class == msg->qos.txtp.traffic_class) { + msg->qos = tmp_qos; pr_info("(%s) trying to get a CBR shortcut\n", client->dev->name); } else @@ -1521,8 +1615,10 @@ static void __exit atm_mpoa_cleanup(void) mpc = tmp; } + mutex_lock(&qos_mutex); qos = qos_head; qos_head = NULL; + mutex_unlock(&qos_mutex); while (qos != NULL) { nextqos = qos->next; dprintk("freeing qos entry %p\n", qos); diff --git a/net/atm/mpc.h b/net/atm/mpc.h index 454abd07651a..0536946989ad 100644 --- a/net/atm/mpc.h +++ b/net/atm/mpc.h @@ -47,8 +47,9 @@ struct atm_mpoa_qos { /* MPOA QoS operations */ struct atm_mpoa_qos *atm_mpoa_add_qos(__be32 dst_ip, struct atm_qos *qos); -struct atm_mpoa_qos *atm_mpoa_search_qos(__be32 dst_ip); +bool atm_mpoa_get_qos(__be32 dst_ip, struct atm_qos *out); int atm_mpoa_delete_qos(struct atm_mpoa_qos *qos); +int atm_mpoa_delete_qos_by_ip(__be32 dst_ip); /* Display QoS entries. This is for the procfs */ struct seq_file; diff --git a/net/atm/mpoa_caches.c b/net/atm/mpoa_caches.c index f7a2f0e41105..30c5b10ee562 100644 --- a/net/atm/mpoa_caches.c +++ b/net/atm/mpoa_caches.c @@ -132,7 +132,6 @@ static in_cache_entry *in_cache_add_entry(__be32 dst_ip, static int cache_hit(in_cache_entry *entry, struct mpoa_client *mpc) { - struct atm_mpoa_qos *qos; struct k_message msg; entry->count++; @@ -144,9 +143,8 @@ static int cache_hit(in_cache_entry *entry, struct mpoa_client *mpc) msg.type = SND_MPOA_RES_RQST; msg.content.in_info = entry->ctrl_info; memcpy(msg.MPS_ctrl, mpc->mps_ctrl_addr, ATM_ESA_LEN); - qos = atm_mpoa_search_qos(entry->ctrl_info.in_dst_ip); - if (qos != NULL) - msg.qos = qos->qos; + if (!atm_mpoa_get_qos(entry->ctrl_info.in_dst_ip, &msg.qos)) + memset(&msg.qos, 0, sizeof(msg.qos)); msg_to_mpoad(&msg, mpc); entry->reply_wait = ktime_get_seconds(); entry->entry_state = INGRESS_RESOLVING; @@ -167,9 +165,8 @@ static int cache_hit(in_cache_entry *entry, struct mpoa_client *mpc) msg.type = SND_MPOA_RES_RQST; memcpy(msg.MPS_ctrl, mpc->mps_ctrl_addr, ATM_ESA_LEN); msg.content.in_info = entry->ctrl_info; - qos = atm_mpoa_search_qos(entry->ctrl_info.in_dst_ip); - if (qos != NULL) - msg.qos = qos->qos; + if (!atm_mpoa_get_qos(entry->ctrl_info.in_dst_ip, &msg.qos)) + memset(&msg.qos, 0, sizeof(msg.qos)); msg_to_mpoad(&msg, mpc); entry->reply_wait = ktime_get_seconds(); } @@ -249,7 +246,6 @@ static void clear_count_and_expired(struct mpoa_client *client) static void check_resolving_entries(struct mpoa_client *client) { - struct atm_mpoa_qos *qos; in_cache_entry *entry; time64_t now; struct k_message msg; @@ -283,9 +279,8 @@ static void check_resolving_entries(struct mpoa_client *client) msg.type = SND_MPOA_RES_RTRY; memcpy(msg.MPS_ctrl, client->mps_ctrl_addr, ATM_ESA_LEN); msg.content.in_info = entry->ctrl_info; - qos = atm_mpoa_search_qos(entry->ctrl_info.in_dst_ip); - if (qos != NULL) - msg.qos = qos->qos; + if (!atm_mpoa_get_qos(entry->ctrl_info.in_dst_ip, &msg.qos)) + memset(&msg.qos, 0, sizeof(msg.qos)); msg_to_mpoad(&msg, client); entry->reply_wait = ktime_get_seconds(); } diff --git a/net/atm/mpoa_proc.c b/net/atm/mpoa_proc.c index aaf64b953915..600075c6022e 100644 --- a/net/atm/mpoa_proc.c +++ b/net/atm/mpoa_proc.c @@ -254,7 +254,7 @@ static int parse_qos(const char *buff) if (sscanf(buff, "del %hhu.%hhu.%hhu.%hhu", ip, ip+1, ip+2, ip+3) == 4) { ipaddr = *(__be32 *)ip; - return atm_mpoa_delete_qos(atm_mpoa_search_qos(ipaddr)); + return atm_mpoa_delete_qos_by_ip(ipaddr); } if (sscanf(buff, "add %hhu.%hhu.%hhu.%hhu tx=%d,%d rx=tx", -- 2.39.5 (Apple Git-154)

4 days, 20 hours

2
1
0 0

[PATCH v2] xfs: Fix a memory leak in xfs_buf_item_init()

by Haoxiang Li

xfs_buf_item_get_format() may allocate memory for bip->bli_formats, free the memory in the error path. Fixes: c3d5f0c2fb85 ("xfs: complain if anyone tries to create a too-large buffer log item") Cc: stable(a)vger.kernel.org Signed-off-by: Haoxiang Li <lihaoxiang(a)isrc.iscas.ac.cn> Reviewed-by: Christoph Hellwig <hch(a)lst.de> --- Changes in v2: - Modify the patch subject. Thanks, Christoph! --- fs/xfs/xfs_buf_item.c | 1 + 1 file changed, 1 insertion(+) diff --git a/fs/xfs/xfs_buf_item.c b/fs/xfs/xfs_buf_item.c index 8d85b5eee444..f4c5be67826e 100644 --- a/fs/xfs/xfs_buf_item.c +++ b/fs/xfs/xfs_buf_item.c @@ -896,6 +896,7 @@ xfs_buf_item_init( map_size = DIV_ROUND_UP(chunks, NBWORD); if (map_size > XFS_BLF_DATAMAP_SIZE) { + xfs_buf_item_free_format(bip); kmem_cache_free(xfs_buf_item_cache, bip); xfs_err(mp, "buffer item dirty bitmap (%u uints) too small to reflect %u bytes!", -- 2.25.1

4 days, 21 hours

2
1
0 0

[PATCH] usb: dwc3: of-simple: fix clock resource leak in dwc3_of_simple_probe

by Miaoqian Lin

When clk_bulk_prepare_enable() fails, the error path jumps to err_resetc_assert, skipping clk_bulk_put_all() and leaking the clock references acquired by clk_bulk_get_all(). Add err_clk_put_all label to properly release clock resources in all error paths. Found via static analysis and code review. Fixes: c0c61471ef86 ("usb: dwc3: of-simple: Convert to bulk clk API") Cc: stable(a)vger.kernel.org Signed-off-by: Miaoqian Lin <linmq006(a)gmail.com> --- drivers/usb/dwc3/dwc3-of-simple.c | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) diff --git a/drivers/usb/dwc3/dwc3-of-simple.c b/drivers/usb/dwc3/dwc3-of-simple.c index a4954a21be93..c116143335d9 100644 --- a/drivers/usb/dwc3/dwc3-of-simple.c +++ b/drivers/usb/dwc3/dwc3-of-simple.c @@ -70,11 +70,11 @@ static int dwc3_of_simple_probe(struct platform_device *pdev) simple->num_clocks = ret; ret = clk_bulk_prepare_enable(simple->num_clocks, simple->clks); if (ret) - goto err_resetc_assert; + goto err_clk_put_all; ret = of_platform_populate(np, NULL, NULL, dev); if (ret) - goto err_clk_put; + goto err_clk_disable; pm_runtime_set_active(dev); pm_runtime_enable(dev); @@ -82,8 +82,9 @@ static int dwc3_of_simple_probe(struct platform_device *pdev) return 0; -err_clk_put: +err_clk_disable: clk_bulk_disable_unprepare(simple->num_clocks, simple->clks); +err_clk_put_all: clk_bulk_put_all(simple->num_clocks, simple->clks); err_resetc_assert: -- 2.25.1

4 days, 23 hours

1
0
0 0

[PATCH 6.6.y] xfrm: state: fix out-of-bounds read during lookup

by Rajani Kantha

From: Florian Westphal <fw(a)strlen.de> [ Upstream commit e952837f3ddb0ff726d5b582aa1aad9aa38d024d ] lookup and resize can run in parallel. The xfrm_state_hash_generation seqlock ensures a retry, but the hash functions can observe a hmask value that is too large for the new hlist array. rehash does: rcu_assign_pointer(net->xfrm.state_bydst, ndst) [..] net->xfrm.state_hmask = nhashmask; While state lookup does: h = xfrm_dst_hash(net, daddr, saddr, tmpl->reqid, encap_family); hlist_for_each_entry_rcu(x, net->xfrm.state_bydst + h, bydst) { This is only safe in case the update to state_bydst is larger than net->xfrm.xfrm_state_hmask (or if the lookup function gets serialized via state spinlock again). Fix this by prefetching state_hmask and the associated pointers. The xfrm_state_hash_generation seqlock retry will ensure that the pointer and the hmask will be consistent. The existing helpers, like xfrm_dst_hash(), are now unsafe for RCU side, add lockdep assertions to document that they are only safe for insert side. xfrm_state_lookup_byaddr() uses the spinlock rather than RCU. AFAICS this is an oversight from back when state lookup was converted to RCU, this lock should be replaced with RCU in a future patch. Reported-by: syzbot+5f9f31cb7d985f584d8e(a)syzkaller.appspotmail.com Closes: https://lore.kernel.org/netdev/CACT4Y+azwfrE3uz6A5ZErov5YN2LYBN5KrsymBerT36… Diagnosed-by: Dmitry Vyukov <dvyukov(a)google.com> Fixes: c2f672fc9464 ("xfrm: state lookup can be lockless") Signed-off-by: Florian Westphal <fw(a)strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert(a)secunet.com> [ Minor context change fixed ] Signed-off-by: Rajani Kantha <681739313(a)139.com> --- net/xfrm/xfrm_state.c | 84 +++++++++++++++++++++++++++++++++---------- 1 file changed, 66 insertions(+), 18 deletions(-) diff --git a/net/xfrm/xfrm_state.c b/net/xfrm/xfrm_state.c index ded559f55767..6512251d3b7a 100644 --- a/net/xfrm/xfrm_state.c +++ b/net/xfrm/xfrm_state.c @@ -34,6 +34,8 @@ #define xfrm_state_deref_prot(table, net) \ rcu_dereference_protected((table), lockdep_is_held(&(net)->xfrm.xfrm_state_lock)) +#define xfrm_state_deref_check(table, net) \ + rcu_dereference_check((table), lockdep_is_held(&(net)->xfrm.xfrm_state_lock)) static void xfrm_state_gc_task(struct work_struct *work); @@ -62,6 +64,8 @@ static inline unsigned int xfrm_dst_hash(struct net *net, u32 reqid, unsigned short family) { + lockdep_assert_held(&net->xfrm.xfrm_state_lock); + return __xfrm_dst_hash(daddr, saddr, reqid, family, net->xfrm.state_hmask); } @@ -70,6 +74,8 @@ static inline unsigned int xfrm_src_hash(struct net *net, const xfrm_address_t *saddr, unsigned short family) { + lockdep_assert_held(&net->xfrm.xfrm_state_lock); + return __xfrm_src_hash(daddr, saddr, family, net->xfrm.state_hmask); } @@ -77,11 +83,15 @@ static inline unsigned int xfrm_spi_hash(struct net *net, const xfrm_address_t *daddr, __be32 spi, u8 proto, unsigned short family) { + lockdep_assert_held(&net->xfrm.xfrm_state_lock); + return __xfrm_spi_hash(daddr, spi, proto, family, net->xfrm.state_hmask); } static unsigned int xfrm_seq_hash(struct net *net, u32 seq) { + lockdep_assert_held(&net->xfrm.xfrm_state_lock); + return __xfrm_seq_hash(seq, net->xfrm.state_hmask); } @@ -1029,16 +1039,38 @@ xfrm_init_tempstate(struct xfrm_state *x, const struct flowi *fl, x->props.family = tmpl->encap_family; } -static struct xfrm_state *__xfrm_state_lookup_all(struct net *net, u32 mark, +struct xfrm_hash_state_ptrs { + const struct hlist_head *bydst; + const struct hlist_head *bysrc; + const struct hlist_head *byspi; + unsigned int hmask; +}; + +static void xfrm_hash_ptrs_get(const struct net *net, struct xfrm_hash_state_ptrs *ptrs) +{ + unsigned int sequence; + + do { + sequence = read_seqcount_begin(&net->xfrm.xfrm_state_hash_generation); + + ptrs->bydst = xfrm_state_deref_check(net->xfrm.state_bydst, net); + ptrs->bysrc = xfrm_state_deref_check(net->xfrm.state_bysrc, net); + ptrs->byspi = xfrm_state_deref_check(net->xfrm.state_byspi, net); + ptrs->hmask = net->xfrm.state_hmask; + } while (read_seqcount_retry(&net->xfrm.xfrm_state_hash_generation, sequence)); +} + +static struct xfrm_state *__xfrm_state_lookup_all(const struct xfrm_hash_state_ptrs *state_ptrs, + u32 mark, const xfrm_address_t *daddr, __be32 spi, u8 proto, unsigned short family, struct xfrm_dev_offload *xdo) { - unsigned int h = xfrm_spi_hash(net, daddr, spi, proto, family); + unsigned int h = __xfrm_spi_hash(daddr, spi, proto, family, state_ptrs->hmask); struct xfrm_state *x; - hlist_for_each_entry_rcu(x, net->xfrm.state_byspi + h, byspi) { + hlist_for_each_entry_rcu(x, state_ptrs->byspi + h, byspi) { #ifdef CONFIG_XFRM_OFFLOAD if (xdo->type == XFRM_DEV_OFFLOAD_PACKET) { if (x->xso.type != XFRM_DEV_OFFLOAD_PACKET) @@ -1072,15 +1104,16 @@ static struct xfrm_state *__xfrm_state_lookup_all(struct net *net, u32 mark, return NULL; } -static struct xfrm_state *__xfrm_state_lookup(struct net *net, u32 mark, +static struct xfrm_state *__xfrm_state_lookup(const struct xfrm_hash_state_ptrs *state_ptrs, + u32 mark, const xfrm_address_t *daddr, __be32 spi, u8 proto, unsigned short family) { - unsigned int h = xfrm_spi_hash(net, daddr, spi, proto, family); + unsigned int h = __xfrm_spi_hash(daddr, spi, proto, family, state_ptrs->hmask); struct xfrm_state *x; - hlist_for_each_entry_rcu(x, net->xfrm.state_byspi + h, byspi) { + hlist_for_each_entry_rcu(x, state_ptrs->byspi + h, byspi) { if (x->props.family != family || x->id.spi != spi || x->id.proto != proto || @@ -1097,15 +1130,16 @@ static struct xfrm_state *__xfrm_state_lookup(struct net *net, u32 mark, return NULL; } -static struct xfrm_state *__xfrm_state_lookup_byaddr(struct net *net, u32 mark, +static struct xfrm_state *__xfrm_state_lookup_byaddr(const struct xfrm_hash_state_ptrs *state_ptrs, + u32 mark, const xfrm_address_t *daddr, const xfrm_address_t *saddr, u8 proto, unsigned short family) { - unsigned int h = xfrm_src_hash(net, daddr, saddr, family); + unsigned int h = __xfrm_src_hash(daddr, saddr, family, state_ptrs->hmask); struct xfrm_state *x; - hlist_for_each_entry_rcu(x, net->xfrm.state_bysrc + h, bysrc) { + hlist_for_each_entry_rcu(x, state_ptrs->bysrc + h, bysrc) { if (x->props.family != family || x->id.proto != proto || !xfrm_addr_equal(&x->id.daddr, daddr, family) || @@ -1125,14 +1159,17 @@ static struct xfrm_state *__xfrm_state_lookup_byaddr(struct net *net, u32 mark, static inline struct xfrm_state * __xfrm_state_locate(struct xfrm_state *x, int use_spi, int family) { + struct xfrm_hash_state_ptrs state_ptrs; struct net *net = xs_net(x); u32 mark = x->mark.v & x->mark.m; + xfrm_hash_ptrs_get(net, &state_ptrs); + if (use_spi) - return __xfrm_state_lookup(net, mark, &x->id.daddr, + return __xfrm_state_lookup(&state_ptrs, mark, &x->id.daddr, x->id.spi, x->id.proto, family); else - return __xfrm_state_lookup_byaddr(net, mark, + return __xfrm_state_lookup_byaddr(&state_ptrs, mark, &x->id.daddr, &x->props.saddr, x->id.proto, family); @@ -1195,6 +1232,7 @@ xfrm_state_find(const xfrm_address_t *daddr, const xfrm_address_t *saddr, unsigned short family, u32 if_id) { static xfrm_address_t saddr_wildcard = { }; + struct xfrm_hash_state_ptrs state_ptrs; struct net *net = xp_net(pol); unsigned int h, h_wildcard; struct xfrm_state *x, *x0, *to_put; @@ -1211,8 +1249,10 @@ xfrm_state_find(const xfrm_address_t *daddr, const xfrm_address_t *saddr, sequence = read_seqcount_begin(&net->xfrm.xfrm_state_hash_generation); rcu_read_lock(); - h = xfrm_dst_hash(net, daddr, saddr, tmpl->reqid, encap_family); - hlist_for_each_entry_rcu(x, net->xfrm.state_bydst + h, bydst) { + xfrm_hash_ptrs_get(net, &state_ptrs); + + h = __xfrm_dst_hash(daddr, saddr, tmpl->reqid, encap_family, state_ptrs.hmask); + hlist_for_each_entry_rcu(x, state_ptrs.bydst + h, bydst) { #ifdef CONFIG_XFRM_OFFLOAD if (pol->xdo.type == XFRM_DEV_OFFLOAD_PACKET) { if (x->xso.type != XFRM_DEV_OFFLOAD_PACKET) @@ -1245,8 +1285,9 @@ xfrm_state_find(const xfrm_address_t *daddr, const xfrm_address_t *saddr, if (best || acquire_in_progress) goto found; - h_wildcard = xfrm_dst_hash(net, daddr, &saddr_wildcard, tmpl->reqid, encap_family); - hlist_for_each_entry_rcu(x, net->xfrm.state_bydst + h_wildcard, bydst) { + h_wildcard = __xfrm_dst_hash(daddr, &saddr_wildcard, tmpl->reqid, + encap_family, state_ptrs.hmask); + hlist_for_each_entry_rcu(x, state_ptrs.bydst + h_wildcard, bydst) { #ifdef CONFIG_XFRM_OFFLOAD if (pol->xdo.type == XFRM_DEV_OFFLOAD_PACKET) { if (x->xso.type != XFRM_DEV_OFFLOAD_PACKET) @@ -1281,7 +1322,7 @@ xfrm_state_find(const xfrm_address_t *daddr, const xfrm_address_t *saddr, x = best; if (!x && !error && !acquire_in_progress) { if (tmpl->id.spi && - (x0 = __xfrm_state_lookup_all(net, mark, daddr, + (x0 = __xfrm_state_lookup_all(&state_ptrs, mark, daddr, tmpl->id.spi, tmpl->id.proto, encap_family, &pol->xdo)) != NULL) { @@ -2036,10 +2077,13 @@ struct xfrm_state * xfrm_state_lookup(struct net *net, u32 mark, const xfrm_address_t *daddr, __be32 spi, u8 proto, unsigned short family) { + struct xfrm_hash_state_ptrs state_ptrs; struct xfrm_state *x; rcu_read_lock(); - x = __xfrm_state_lookup(net, mark, daddr, spi, proto, family); + xfrm_hash_ptrs_get(net, &state_ptrs); + + x = __xfrm_state_lookup(&state_ptrs, mark, daddr, spi, proto, family); rcu_read_unlock(); return x; } @@ -2050,10 +2094,14 @@ xfrm_state_lookup_byaddr(struct net *net, u32 mark, const xfrm_address_t *daddr, const xfrm_address_t *saddr, u8 proto, unsigned short family) { + struct xfrm_hash_state_ptrs state_ptrs; struct xfrm_state *x; spin_lock_bh(&net->xfrm.xfrm_state_lock); - x = __xfrm_state_lookup_byaddr(net, mark, daddr, saddr, proto, family); + + xfrm_hash_ptrs_get(net, &state_ptrs); + + x = __xfrm_state_lookup_byaddr(&state_ptrs, mark, daddr, saddr, proto, family); spin_unlock_bh(&net->xfrm.xfrm_state_lock); return x; } -- 2.17.1

4 days, 23 hours

1
0
0 0

Request to backport net/lan743x crashkernel fix to 6.1, 6.6, and 6.12 branches.

by Zhang, Liyin (CN)

Hi, The following patch has already been merged into mainline (master) and the linux-6.12.y stable branch: net: lan743x: Allocate rings outside ZONE_DMA (commit 8a8f3f4991761a70834fe6719d09e9fd338a766e) https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id… https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=… I believe this patch should be backported. Our customer is experiencing an issue where a Microchip LAN7430 NIC fails to work in a crashkernel environment. Because the driver uses the ZONE_DMA flag, memory allocation fails when crashkernel reserves memory — preventing the card from functioning, especially when they try to transfer the VMcore file over network. We are using older kernel versions and request that this fix be applied to the following branches: linux-6.1.y linux-6.6.y linux-6.12.y Could you please help backport this fix? Thank you for your time and support. Best regards, Liyin

5 days

1
1
0 0

[PATCH v1 1/4] mm/hugetlb: fix hugetlb_pmd_shared()

by David Hildenbrand (Red Hat)

We switched from (wrongly) using the page count to an independent shared count. Now, shared page tables have a refcount of 1 (excluding speculative references) and instead use ptdesc->pt_share_count to identify sharing. We didn't convert hugetlb_pmd_shared(), so right now, we would never detect a shared PMD table as such, because sharing/unsharing no longer touches the refcount of a PMD table. Page migration, like mbind() or migrate_pages() would allow for migrating folios mapped into such shared PMD tables, even though the folios are not exclusive. In smaps we would account them as "private" although they are "shared", and we would be wrongly setting the PM_MMAP_EXCLUSIVE in the pagemap interface. Fix it by properly using ptdesc_pmd_is_shared() in hugetlb_pmd_shared(). Fixes: 59d9094df3d7 ("mm: hugetlb: independent PMD page table shared count") Cc: <stable(a)vger.kernel.org> Cc: Liu Shixin <liushixin2(a)huawei.com> Signed-off-by: David Hildenbrand (Red Hat) <david(a)kernel.org> --- include/linux/hugetlb.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/include/linux/hugetlb.h b/include/linux/hugetlb.h index 019a1c5281e4e..03c8725efa289 100644 --- a/include/linux/hugetlb.h +++ b/include/linux/hugetlb.h @@ -1326,7 +1326,7 @@ static inline __init void hugetlb_cma_reserve(int order) #ifdef CONFIG_HUGETLB_PMD_PAGE_TABLE_SHARING static inline bool hugetlb_pmd_shared(pte_t *pte) { - return page_count(virt_to_page(pte)) > 1; + return ptdesc_pmd_is_shared(virt_to_ptdesc(pte)); } #else static inline bool hugetlb_pmd_shared(pte_t *pte) -- 2.52.0

5 days

7
9
0 0

Great Original Shirts

by Spencer

5 days, 1 hour

1
0
0 0

[PATCH v2 0/2] iommu/vt-d: Skip dev-iotlb flush for inaccessible PCIe device

by Jinhui Guo

Hi, all We hit hard-lockups when the Intel IOMMU waits indefinitely for an ATS invalidation that cannot complete, especially under GDR high-load conditions. 1. Hard-lock when a passthrough PCIe NIC with ATS enabled link-down in Intel IOMMU non-scalable mode. Two scenarios exist: NIC link-down with an explicit link-down event and link-down without any event. a) NIC link-down with an explicit link-dow event. Call Trace: qi_submit_sync qi_flush_dev_iotlb __context_flush_dev_iotlb.part.0 domain_context_clear_one_cb pci_for_each_dma_alias device_block_translation blocking_domain_attach_dev iommu_deinit_device __iommu_group_remove_device iommu_release_device iommu_bus_notifier blocking_notifier_call_chain bus_notify device_del pci_remove_bus_device pci_stop_and_remove_bus_device pciehp_unconfigure_device pciehp_disable_slot pciehp_handle_presence_or_link_change pciehp_ist b) NIC link-down without an event - hard-lock on VM destroy. Call Trace: qi_submit_sync qi_flush_dev_iotlb __context_flush_dev_iotlb.part.0 domain_context_clear_one_cb pci_for_each_dma_alias device_block_translation blocking_domain_attach_dev __iommu_attach_device __iommu_device_set_domain __iommu_group_set_domain_internal iommu_detach_group vfio_iommu_type1_detach_group vfio_group_detach_container vfio_group_fops_release __fput 2. Hard-lock when a passthrough PCIe NIC with ATS enabled link-down in Intel IOMMU scalable mode; NIC link-down without an event hard-locks on VM destroy. Call Trace: qi_submit_sync qi_flush_dev_iotlb intel_pasid_tear_down_entry device_block_translation blocking_domain_attach_dev __iommu_attach_device __iommu_device_set_domain __iommu_group_set_domain_internal iommu_detach_group vfio_iommu_type1_detach_group vfio_group_detach_container vfio_group_fops_release __fput Fix both issues with two patches: 1. Skip dev-IOTLB flush for inaccessible devices in __context_flush_dev_iotlb() using pci_device_is_present(). 2. Use pci_device_is_present() instead of pci_dev_is_disconnected() to decide when to skip ATS invalidation in devtlb_invalidation_with_pasid(). Best Regards, Jinhui --- v1: https://lore.kernel.org/all/20251210171431.1589-1-guojinhui.liam@bytedance.… Changelog in v1 -> v2 (suggested by Baolu Lu) - Simplify the pci_device_is_present() check in __context_flush_dev_iotlb(). - Add Cc: stable(a)vger.kernel.org to both patches. Jinhui Guo (2): iommu/vt-d: Skip dev-iotlb flush for inaccessible PCIe device without scalable mode iommu/vt-d: Flush dev-IOTLB only when PCIe device is accessible in scalable mode drivers/iommu/intel/pasid.c | 11 ++++++++++- 1 file changed, 10 insertions(+), 1 deletion(-) -- 2.20.1

5 days, 2 hours

1
2
0 0

2025

2024

2023

2022

2021

2020

2019

2018

2017

Linux-stable-mirror