kcov: don't generate a warning on vm_insert_page()'s failure
PATCH v3 1/1 kcov: don't generate a warning on vm_insert_page()'s failure
vm_insert_page()'s failure is not an unexpected condition, so don't do WARN_ONCE() in such a case.
Instead, print a kernel message and just return an error code.
---
PATCH v3:
* Adjusted the patch format.
PATCH v2:
* Added a newline at the end of pr_warn_once().
https://lore.kernel.org/all/[email protected]/
PATCH v1:
https://lore.kernel.org/all/[email protected]/
--- kernel/kcov.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-)
kernel/kcov.c +5 -2
@@ -475,8 +475,11 @@ static int kcov_mmap(struct file *filep, struct vm_area_struct *vma) vma->vm_flags |=VM_DONTEXPAND ;for (off =0 ; off < size; off +=PAGE_SIZE ) { page =vmalloc_to_page (kcov->area + off); -if (vm_insert_page (vma, vma->vm_start + off, page)) -WARN_ONCE (1 ,"vm_insert_page() failed" ); + res =vm_insert_page (vma, vma->vm_start + off, page); +if (res) { +pr_warn_once ("kcov: vm_insert_page() failed\n" ); +return res; + } }return 0 ; exit:
Signature · 1 line
2.35.1.1094.g7c7d902a7c-goog
Quoted 13 lines · On Fri, 1 Apr 2022 18:25:12 +0000 Aleksandr Nogikh <[email protected]> wrote:
vm_insert_page()'s failure is not an unexpected condition, so don't do
WARN_ONCE()in such a case.Instead, print a kernel message and just return an error code.
...
kernel/kcov.c +0
-0@@ -475,8 +475,11 @@ static int kcov_mmap(struct file *filep, struct vm_area_struct *vma) vma->vm_flags |=VM_DONTEXPAND ;for (off =0 ; off < size; off +=PAGE_SIZE ) { page =vmalloc_to_page (kcov->area + off);
kernel/kcov.c static int kcov_mmap(struct file *filep, struct vm_area_struct *vma) · line 485
Quoted 10 lines
kernel/kcov.c +5
-2-if (vm_insert_page (vma, vma->vm_start + off, page)) -WARN_ONCE (1 ,"vm_insert_page() failed" ); + res =vm_insert_page (vma, vma->vm_start + off, page); +if (res) { +pr_warn_once ("kcov: vm_insert_page() failed\n" ); +return res; + } }return 0 ; exit:
Can you explain the rationale here? If vm_insert_page() failure is an expected condition, why warn at all?
I'm struggling to understand why a condition is worth a printk, but not a WARN.
Some explanation of what leads to the vm_insert_page() failure would have been helpful.
Quoted 4 lines · On Fri, 1 Apr 2022 18:25:12 +0000 Aleksandr Nogikh <[email protected]> wrote:
vm_insert_page()'s failure is not an unexpected condition, so don't do
WARN_ONCE()in such a case.Instead, print a kernel message and just return an error code.
(hm, I thought I asked this before but I can't find it)
Under what circumstances will this failure occur?
Why do we emit a message at all? What action can the user take upon seeing the message?
Do we have a Fixes: for this?
From the info provided thus far I'm unable to determine whether a -stable backport is needed. What are your thoughts on this?
Quoted 10 lines · Andrew Morton
Quoted 4 lines · On Fri, 1 Apr 2022 18:25:12 +0000 Aleksandr Nogikh <[email protected]> wrote:
vm_insert_page()'s failure is not an unexpected condition, so don't do
WARN_ONCE()in such a case.Instead, print a kernel message and just return an error code.
(hm, I thought I asked this before but I can't find it)
Under what circumstances will this failure occur?
It looks like syzbot was able to generate an OOM situation:
| [ 599.515700][T23028] oom-kill:constraint=CONSTRAINT_MEMCG,nodemask=(null),cpuset=syz1,mems_allowed=0-1,oom_memcg=/syz1,task_memcg=/syz1,task=syz-executor.1,pid=23028,uid=0 | [ 599.537757][T23028] Memory cgroup out of memory: Killed process 23028 (syz-executor.1) total-vm:56816kB, anon-rss:436kB, file-rss:8888kB, shmem-rss:48kB, UID:0 pgtables:88kB oom_score_adj:1000 | [ 599.615664][T23028] ------------[ cut here ]------------ | [ 599.652858][T23028] vm_insert_page() failed | [ 599.662598][T23028] WARNING: CPU: 0 PID: 23028 at kernel/kcov.c:479 kcov_mmap+0xbe/0xe0 | [ 599.900577][T23028] Modules linked in: | [ 599.904480][T23028] CPU: 1 PID: 23028 Comm: syz-executor.1 Tainted: G W 5.17.0-syzkaller-12964-gccaff3d56acc #0 | [ 599.956099][T23028] Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 | [ 600.092674][T23028] RIP: 0010:kcov_mmap+0xbe/0xe0 | [ 600.097559][T23028] Code: 48 81 c3 00 10 00 00 49 39 dc 77 c9 31 c0 5b 5d 41 5c 41 5d 41 5e c3 48 c7 c7 e9 4a 5b 8b c6 05 5a fc 28 0c 01 e8 bd c6 a0 07 <0f> 0b eb d2 4c 89 f7 e8 66 28 e8 07 b8 ea ff ff ff eb d1 66 66 2e | [ 600.117319][T23028] RSP: 0018:ffffc9000c1cfc30 EFLAGS: 00010282 | [ 600.135794][T23028] RAX: 0000000000000000 RBX: 0000000000000000 RCX: 0000000000000000 | [ 600.163986][T23028] RDX: ffff888051f40000 RSI: ffffffff815fce18 RDI: fffff52001839f78 | [ 600.188615][T23028] RBP: ffff88804fc6e210 R08: 0000000000000000 R09: 0000000000000000 | [ 600.196616][T23028] R10: ffffffff815f77ee R11: 0000000000000000 R12: 0000000000200000 | [ 600.214229][T23028] R13: ffff8880646c2500 R14: ffff8880646c2508 R15: ffff88804fc6e260 | [ 600.252864][T23028] FS: 00005555570e4400(0000) GS:ffff8880b9c00000(0000) knlGS:0000000000000000 | [ 600.283249][T23028] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 | [ 600.335749][T23028] CR2: 0000001b2c436000 CR3: 000000004ef16000 CR4: 00000000003506f0 | [ 600.390781][T23028] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 | [ 600.430312][T23028] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000600 | [ 600.441698][T23028] Call Trace: | [ 600.447877][T23028] <TASK> | [ 600.451890][T23028] mmap_region+0xba5/0x14a0 | [ 600.486043][T23028] do_mmap+0x863/0xfa0 | [ 600.490544][T23028] vm_mmap_pgoff+0x1b7/0x290 | [ 600.505607][T23028] ksys_mmap_pgoff+0x40d/0x5a0 | [ 600.522165][T23028] do_syscall_64+0x35/0x80 | [ 600.526655][T23028] entry_SYSCALL_64_after_hwframe+0x44/0xae | [ 600.532936][T23028] RIP: 0033:0x7f5be4889092 | [ 600.537407][T23028] Code: 00 00 00 00 00 0f 1f 00 41 f7 c1 ff 0f 00 00 75 27 55 48 89 fd 53 89 cb 48 85 ff 74 3b 41 89 da 48 89 ef b8 09 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 66 5b 5d c3 0f 1f 00 48 c7 c0 b8 ff ff ff 64 | [ 600.560042][T23028] RSP: 002b:00007fffde76b318 EFLAGS: 00000246 ORIG_RAX: 0000000000000009 | [ 600.569079][T23028] RAX: ffffffffffffffda RBX: 0000000000000001 RCX: 00007f5be4889092 | [ 600.577107][T23028] RDX: 0000000000000003 RSI: 0000000000200000 RDI: 0000000000000000 | [ 600.587064][T23028] RBP: 0000000000000000 R08: 00000000000000db R09: 0000000000000000 | [ 600.596119][T23028] R10: 0000000000000001 R11: 0000000000000246 R12: 00007f5be499c1dc | [ 600.604977][T23028] R13: 0000000000000003 R14: 00007f5be499c1d0 R15: 0000000000000032 | [ 600.613026][T23028] </TASK> | [ 600.616066][T23028] Kernel panic - not syncing: panic_on_warn set ...
Quoted 2 lines
Why do we emit a message at all? What action can the user take upon seeing the message?
The message is mainly for the benefit of the test log, in this case the fuzzer's log so that humans inspecting the log can figure out what was going on. KCOV is a testing tool, so I think being a little more chatty when KCOV unexpectedly is about to fail will save someone debugging time.
We don't want the WARN, because it's not a kernel bug that syzbot should report, and failure can happen if the fuzzer tries hard enough (as above).
Quoted 1 line
Do we have a Fixes: for this?
The WARN was moved with b3d7fe86fbd0 ("kcov: properly handle subsequent mmap calls"), so that'd be the only commit a backport would cleanly apply to.
Quoted 2 lines
From the info provided thus far I'm unable to determine whether a -stable backport is needed. What are your thoughts on this?
The main problem is it only makes fuzzers try to report this as a bug (which it is not). Backporting to kernels that have b3d7fe86fbd0 would be reasonable, but wouldn't bother creating backports for older kernels.
Thanks,
-- Marco
Marco, thank you very much for these answers!
I was unfortunately a bit overloaded lately, so was postponing a reply (given that the patch is very good to have, but not urgent).
Quoted 77 lines · Marco Elver
Quoted 10 lines · Andrew Morton
Quoted 4 lines · On Fri, 1 Apr 2022 18:25:12 +0000 Aleksandr Nogikh <[email protected]> wrote:
vm_insert_page()'s failure is not an unexpected condition, so don't do
WARN_ONCE()in such a case.Instead, print a kernel message and just return an error code.
(hm, I thought I asked this before but I can't find it)
Under what circumstances will this failure occur?
It looks like syzbot was able to generate an OOM situation:
| [ 599.515700][T23028] oom-kill:constraint=CONSTRAINT_MEMCG,nodemask=(null),cpuset=syz1,mems_allowed=0-1,oom_memcg=/syz1,task_memcg=/syz1,task=syz-executor.1,pid=23028,uid=0 | [ 599.537757][T23028] Memory cgroup out of memory: Killed process 23028 (syz-executor.1) total-vm:56816kB, anon-rss:436kB, file-rss:8888kB, shmem-rss:48kB, UID:0 pgtables:88kB oom_score_adj:1000 | [ 599.615664][T23028] ------------[ cut here ]------------ | [ 599.652858][T23028] vm_insert_page() failed | [ 599.662598][T23028] WARNING: CPU: 0 PID: 23028 at kernel/kcov.c:479 kcov_mmap+0xbe/0xe0 | [ 599.900577][T23028] Modules linked in: | [ 599.904480][T23028] CPU: 1 PID: 23028 Comm: syz-executor.1 Tainted: G W 5.17.0-syzkaller-12964-gccaff3d56acc #0 | [ 599.956099][T23028] Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 | [ 600.092674][T23028] RIP: 0010:kcov_mmap+0xbe/0xe0 | [ 600.097559][T23028] Code: 48 81 c3 00 10 00 00 49 39 dc 77 c9 31 c0 5b 5d 41 5c 41 5d 41 5e c3 48 c7 c7 e9 4a 5b 8b c6 05 5a fc 28 0c 01 e8 bd c6 a0 07 <0f> 0b eb d2 4c 89 f7 e8 66 28 e8 07 b8 ea ff ff ff eb d1 66 66 2e | [ 600.117319][T23028] RSP: 0018:ffffc9000c1cfc30 EFLAGS: 00010282 | [ 600.135794][T23028] RAX: 0000000000000000 RBX: 0000000000000000 RCX: 0000000000000000 | [ 600.163986][T23028] RDX: ffff888051f40000 RSI: ffffffff815fce18 RDI: fffff52001839f78 | [ 600.188615][T23028] RBP: ffff88804fc6e210 R08: 0000000000000000 R09: 0000000000000000 | [ 600.196616][T23028] R10: ffffffff815f77ee R11: 0000000000000000 R12: 0000000000200000 | [ 600.214229][T23028] R13: ffff8880646c2500 R14: ffff8880646c2508 R15: ffff88804fc6e260 | [ 600.252864][T23028] FS: 00005555570e4400(0000) GS:ffff8880b9c00000(0000) knlGS:0000000000000000 | [ 600.283249][T23028] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 | [ 600.335749][T23028] CR2: 0000001b2c436000 CR3: 000000004ef16000 CR4: 00000000003506f0 | [ 600.390781][T23028] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 | [ 600.430312][T23028] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000600 | [ 600.441698][T23028] Call Trace: | [ 600.447877][T23028] <TASK> | [ 600.451890][T23028] mmap_region+0xba5/0x14a0 | [ 600.486043][T23028] do_mmap+0x863/0xfa0 | [ 600.490544][T23028] vm_mmap_pgoff+0x1b7/0x290 | [ 600.505607][T23028] ksys_mmap_pgoff+0x40d/0x5a0 | [ 600.522165][T23028] do_syscall_64+0x35/0x80 | [ 600.526655][T23028] entry_SYSCALL_64_after_hwframe+0x44/0xae | [ 600.532936][T23028] RIP: 0033:0x7f5be4889092 | [ 600.537407][T23028] Code: 00 00 00 00 00 0f 1f 00 41 f7 c1 ff 0f 00 00 75 27 55 48 89 fd 53 89 cb 48 85 ff 74 3b 41 89 da 48 89 ef b8 09 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 66 5b 5d c3 0f 1f 00 48 c7 c0 b8 ff ff ff 64 | [ 600.560042][T23028] RSP: 002b:00007fffde76b318 EFLAGS: 00000246 ORIG_RAX: 0000000000000009 | [ 600.569079][T23028] RAX: ffffffffffffffda RBX: 0000000000000001 RCX: 00007f5be4889092 | [ 600.577107][T23028] RDX: 0000000000000003 RSI: 0000000000200000 RDI: 0000000000000000 | [ 600.587064][T23028] RBP: 0000000000000000 R08: 00000000000000db R09: 0000000000000000 | [ 600.596119][T23028] R10: 0000000000000001 R11: 0000000000000246 R12: 00007f5be499c1dc | [ 600.604977][T23028] R13: 0000000000000003 R14: 00007f5be499c1d0 R15: 0000000000000032 | [ 600.613026][T23028] </TASK> | [ 600.616066][T23028] Kernel panic - not syncing: panic_on_warn set ...Quoted 2 lines
Why do we emit a message at all? What action can the user take upon seeing the message?
The message is mainly for the benefit of the test log, in this case the fuzzer's log so that humans inspecting the log can figure out what was going on. KCOV is a testing tool, so I think being a little more chatty when KCOV unexpectedly is about to fail will save someone debugging time.
We don't want the WARN, because it's not a kernel bug that syzbot should report, and failure can happen if the fuzzer tries hard enough (as above).
Quoted 1 line
Do we have a Fixes: for this?
The WARN was moved with
b3d7fe86fbd0("kcov: properly handle subsequent mmap calls"), so that'd be the only commit a backport would cleanly apply to.Quoted 2 lines
From the info provided thus far I'm unable to determine whether a -stable backport is needed. What are your thoughts on this?
Quoted 6 lines
The main problem is it only makes fuzzers try to report this as a bug (which it is not). Backporting to kernels that have
b3d7fe86fbd0would be reasonable, but wouldn't bother creating backports for older kernels.Thanks,
-- Marco
I agree with all these points and don't really have anything to add.