> On Jul 20, 2023, at 10:33 AM, Jeff Layton <jlayton@xxxxxxxxxx> wrote: > > On Wed, 2023-07-19 at 23:46 +0000, Chuck Lever III wrote: >> >>> On Jul 19, 2023, at 5:48 PM, Jeff Layton <jlayton@xxxxxxxxxx> wrote: >>> >>> Hi Chuck, >>> >>> While doing some testing today with pynfs on a branch based on your >>> nfsd-next branch. I'm not sure which test triggers it, but it's one of >>> the v4.0 tests. >> >> I've just started running pynfs on nfsd-next, haven't seen any >> crashes so far. I'll take a dive in tomorrow. >> >> >>> It only takes a few mins before it crashes: >>> >>> Jul 19 19:22:43 kdevops-nfsd kernel: BUG: unable to handle page fault for address: ffffd8442d049108 >>> Jul 19 19:22:43 kdevops-nfsd kernel: #PF: supervisor read access in kernel mode >>> Jul 19 19:22:43 kdevops-nfsd kernel: #PF: error_code(0x0000) - not-present page >>> Jul 19 19:22:43 kdevops-nfsd kernel: PGD 0 P4D 0 >>> Jul 19 19:22:43 kdevops-nfsd kernel: Oops: 0000 [#1] PREEMPT SMP PTI >>> Jul 19 19:22:43 kdevops-nfsd kernel: CPU: 0 PID: 743 Comm: nfsd Not tainted 6.5.0-rc2+ #19 >>> Jul 19 19:22:43 kdevops-nfsd kernel: Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 1.16.2-1.fc38 04/01/2014 >>> Jul 19 19:22:43 kdevops-nfsd kernel: RIP: 0010:kfree+0x4b/0x110 >>> Jul 19 19:22:43 kdevops-nfsd kernel: Code: 80 48 01 d8 0f 82 d8 00 00 00 48 c7 c2 00 00 00 80 48 2b 15 9f d7 fb 00 48 01 d0 48 c1 e8 0c 48 c1 e0 06 48 03 05 7d d7 fb 00 <48> 8b 50 08 48 89 c7 f6 c2 01 0f 85 9f 00 00 > >>> Jul 19 19:22:43 kdevops-nfsd kernel: RSP: 0018:ffffb0858661bdc0 EFLAGS: 00010286 >>> Jul 19 19:22:43 kdevops-nfsd kernel: RAX: ffffd8442d049100 RBX: 0000000081244052 RCX: ffff8a665e003008 >>> Jul 19 19:22:43 kdevops-nfsd kernel: RDX: 0000759a40000000 RSI: ffffffffc0b13a45 RDI: 0000000081244052 >>> Jul 19 19:22:43 kdevops-nfsd kernel: RBP: ffffb0858661be40 R08: ffff8a665e003008 R09: ffffffff8ebdc4ec >>> Jul 19 19:22:43 kdevops-nfsd kernel: R10: 000000000000000e R11: 000000000000001b R12: ffff8a6656f20150 >>> Jul 19 19:22:43 kdevops-nfsd kernel: R13: ffff8a665e003008 R14: ffff8a6656f20100 R15: ffff8a6656f20980 >>> Jul 19 19:22:43 kdevops-nfsd kernel: FS: 0000000000000000(0000) GS:ffff8a67bfc00000(0000) knlGS:0000000000000000 >>> Jul 19 19:22:43 kdevops-nfsd kernel: CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 >>> Jul 19 19:22:43 kdevops-nfsd kernel: CR2: ffffd8442d049108 CR3: 0000000277e1a001 CR4: 0000000000060ef0 >>> Jul 19 19:22:43 kdevops-nfsd kernel: Call Trace: >>> Jul 19 19:22:43 kdevops-nfsd kernel: <TASK> >>> Jul 19 19:22:43 kdevops-nfsd kernel: ? __die+0x1f/0x70 >>> Jul 19 19:22:43 kdevops-nfsd kernel: ? page_fault_oops+0x159/0x450 >>> Jul 19 19:22:43 kdevops-nfsd kernel: ? fixup_exception+0x22/0x310 >>> Jul 19 19:22:43 kdevops-nfsd kernel: ? exc_page_fault+0x157/0x180 >>> Jul 19 19:22:43 kdevops-nfsd kernel: ? asm_exc_page_fault+0x22/0x30 >>> Jul 19 19:22:43 kdevops-nfsd kernel: ? nfsd_cache_lookup+0x3c5/0x770 [nfsd] >>> Jul 19 19:22:43 kdevops-nfsd kernel: ? kfree+0x4b/0x110 >>> Jul 19 19:22:43 kdevops-nfsd kernel: nfsd_cache_lookup+0x3c5/0x770 [nfsd] >>> Jul 19 19:22:43 kdevops-nfsd kernel: ? __pfx_nfsd+0x10/0x10 [nfsd] >>> Jul 19 19:22:43 kdevops-nfsd kernel: nfsd_dispatch+0x62/0x1b0 [nfsd] >>> Jul 19 19:22:43 kdevops-nfsd kernel: svc_process_common+0x3cb/0x6c0 [sunrpc] >>> Jul 19 19:22:43 kdevops-nfsd kernel: ? __pfx_nfsd_dispatch+0x10/0x10 [nfsd] >>> Jul 19 19:22:43 kdevops-nfsd kernel: ? __pfx_nfsd+0x10/0x10 [nfsd] >>> Jul 19 19:22:43 kdevops-nfsd kernel: svc_process+0x12d/0x170 [sunrpc] >>> Jul 19 19:22:43 kdevops-nfsd kernel: nfsd+0xd5/0x1a0 [nfsd] >>> Jul 19 19:22:43 kdevops-nfsd kernel: kthread+0xf3/0x120 >>> Jul 19 19:22:43 kdevops-nfsd kernel: ? __pfx_kthread+0x10/0x10 >>> Jul 19 19:22:43 kdevops-nfsd kernel: ret_from_fork+0x30/0x50 >>> Jul 19 19:22:43 kdevops-nfsd kernel: ? __pfx_kthread+0x10/0x10 >>> Jul 19 19:22:43 kdevops-nfsd kernel: ret_from_fork_asm+0x1b/0x30 >>> Jul 19 19:22:43 kdevops-nfsd kernel: RIP: 0000:0x0 >>> Jul 19 19:22:43 kdevops-nfsd kernel: Code: Unable to access opcode bytes at 0xffffffffffffffd6. >>> Jul 19 19:22:43 kdevops-nfsd kernel: RSP: 0000:0000000000000000 EFLAGS: 00000000 ORIG_RAX: 0000000000000000 >>> Jul 19 19:22:43 kdevops-nfsd kernel: RAX: 0000000000000000 RBX: 0000000000000000 RCX: 0000000000000000 >>> Jul 19 19:22:43 kdevops-nfsd kernel: RDX: 0000000000000000 RSI: 0000000000000000 RDI: 0000000000000000 >>> Jul 19 19:22:43 kdevops-nfsd kernel: RBP: 0000000000000000 R08: 0000000000000000 R09: 0000000000000000 >>> Jul 19 19:22:43 kdevops-nfsd kernel: R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000000000 >>> Jul 19 19:22:43 kdevops-nfsd kernel: R13: 0000000000000000 R14: 0000000000000000 R15: 0000000000000000 >>> Jul 19 19:22:43 kdevops-nfsd kernel: </TASK> >>> Jul 19 19:22:43 kdevops-nfsd kernel: Modules linked in: 9p netfs nls_iso8859_1 nls_cp437 vfat fat kvm_intel joydev kvm cirrus psmouse drm_shmem_helper irqbypass pcspkr virtio_net net_failover failover virtio_balloon > >>> Jul 19 19:22:43 kdevops-nfsd kernel: CR2: ffffd8442d049108 >>> Jul 19 19:22:43 kdevops-nfsd kernel: ---[ end trace 0000000000000000 ]--- >>> Jul 19 19:22:43 kdevops-nfsd kernel: RIP: 0010:kfree+0x4b/0x110 >>> Jul 19 19:22:43 kdevops-nfsd kernel: Code: 80 48 01 d8 0f 82 d8 00 00 00 48 c7 c2 00 00 00 80 48 2b 15 9f d7 fb 00 48 01 d0 48 c1 e8 0c 48 c1 e0 06 48 03 05 7d d7 fb 00 <48> 8b 50 08 48 89 c7 f6 c2 01 0f 85 9f 00 00 > >>> Jul 19 19:22:43 kdevops-nfsd kernel: RSP: 0018:ffffb0858661bdc0 EFLAGS: 00010286 >>> Jul 19 19:22:43 kdevops-nfsd kernel: RAX: ffffd8442d049100 RBX: 0000000081244052 RCX: ffff8a665e003008 >>> Jul 19 19:22:43 kdevops-nfsd kernel: RDX: 0000759a40000000 RSI: ffffffffc0b13a45 RDI: 0000000081244052 >>> Jul 19 19:22:43 kdevops-nfsd kernel: RBP: ffffb0858661be40 R08: ffff8a665e003008 R09: ffffffff8ebdc4ec >>> Jul 19 19:22:43 kdevops-nfsd kernel: R10: 000000000000000e R11: 000000000000001b R12: ffff8a6656f20150 >>> Jul 19 19:22:43 kdevops-nfsd kernel: R13: ffff8a665e003008 R14: ffff8a6656f20100 R15: ffff8a6656f20980 >>> Jul 19 19:22:43 kdevops-nfsd kernel: FS: 0000000000000000(0000) GS:ffff8a67bfc00000(0000) knlGS:0000000000000000 >>> Jul 19 19:22:43 kdevops-nfsd kernel: CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 >>> Jul 19 19:22:43 kdevops-nfsd kernel: CR2: ffffffffffffffd6 CR3: 0000000277e1a001 CR4: 0000000000060ef0 >>> Jul 19 19:22:43 kdevops-nfsd kernel: note: nfsd[743] exited with irqs disabled >>> Jul 19 19:22:43 kdevops-nfsd kernel: note: nfsd[743] exited with preempt_count 1 >>> >>> faddr2line says: >>> >>> $ ./scripts/faddr2line --list fs/nfsd/nfsd.ko nfsd_cache_lookup+0x3c5/0x770 >>> nfsd_cacherep_free at /home/jlayton/git/kdevops/linux/fs/nfsd/nfscache.c:116 >>> 111 } >>> 112 >>> 113 static void nfsd_cacherep_free(struct nfsd_cacherep *rp) >>> 114 { >>> 115 kfree(rp->c_replvec.iov_base); >>>> 116< kmem_cache_free(drc_slab, rp); >>> 117 } >>> 118 >>> 119 static unsigned long >>> 120 nfsd_cacherep_dispose(struct list_head *dispose) >>> 121 { >>> >>> (inlined by) nfsd_reply_cache_free_locked at /home/jlayton/git/kdevops/linux/fs/nfsd/nfscache.c:153 >>> 148 static void >>> 149 nfsd_reply_cache_free_locked(struct nfsd_drc_bucket *b, struct nfsd_cacherep *rp, >>> 150 struct nfsd_net *nn) >>> 151 { >>> 152 nfsd_cacherep_unlink_locked(nn, b, rp); >>>> 153< nfsd_cacherep_free(rp); >>> 154 } >>> 155 >>> 156 static void >>> 157 nfsd_reply_cache_free(struct nfsd_drc_bucket *b, struct nfsd_cacherep *rp, >>> 158 struct nfsd_net *nn) >>> >>> (inlined by) nfsd_cache_lookup at /home/jlayton/git/kdevops/linux/fs/nfsd/nfscache.c:527 >>> 522 nfsd_stats_drc_mem_usage_add(nn, sizeof(*rp)); >>> 523 goto out; >>> 524 >>> 525 found_entry: >>> 526 /* We found a matching entry which is either in progress or done. */ >>>> 527< nfsd_reply_cache_free_locked(NULL, rp, nn); >>> 528 nfsd_stats_rc_hits_inc(); >>> 529 rtn = RC_DROPIT; >>> 530 rp = found; >>> 531 >>> 532 /* Request being processed */ >>> >>> >>> ...and a bisect landed here: >>> >>> >>> 767b1d5badd6eb418e3f91f0cd8fa6d2894ff43a is the first bad commit >>> commit 767b1d5badd6eb418e3f91f0cd8fa6d2894ff43a >>> Author: Chuck Lever <chuck.lever@xxxxxxxxxx> >>> Date: Sun Jul 9 11:45:16 2023 -0400 >>> >>> NFSD: Refactor nfsd_reply_cache_free_locked() >>> >>> To reduce contention on the bucket locks, we must avoid calling >>> kfree() while each bucket lock is held. >>> >>> Start by refactoring nfsd_reply_cache_free_locked() into a helper >>> that removes an entry from the bucket (and must therefore run under >>> the lock) and a second helper that frees the entry (which does not >>> need to hold the lock). >>> >>> For readability, rename the helpers nfsd_cacherep_<verb>. >>> >>> Reviewed-by: Jeff Layton <jlayton@xxxxxxxxxx> >>> Signed-off-by: Chuck Lever <chuck.lever@xxxxxxxxxx> >>> >>> fs/nfsd/nfscache.c | 26 +++++++++++++++++++------- >>> 1 file changed, 19 insertions(+), 7 deletions(-) >>> >>> -- >>> Jeff Layton <jlayton@xxxxxxxxxx> >> > > So far, it seems to be surviving with this patch on top. c_replvec is > part of a union, so you need to ensure it actually holds a kvec. > > > diff --git a/fs/nfsd/nfscache.c b/fs/nfsd/nfscache.c > index 6eb3d7bdfaf3..80621a709510 100644 > --- a/fs/nfsd/nfscache.c > +++ b/fs/nfsd/nfscache.c > @@ -112,7 +112,8 @@ nfsd_cacherep_alloc(struct svc_rqst *rqstp, __wsum csum, > > static void nfsd_cacherep_free(struct nfsd_cacherep *rp) > { > - kfree(rp->c_replvec.iov_base); > + if (rp->c_type == RC_REPLBUFF) > + kfree(rp->c_replvec.iov_base); > kmem_cache_free(drc_slab, rp); > } Agreed, that was an unintended change to the logic introduced by that commit, and it lines up with the RIP. -- Chuck Lever