Re: kern/57198: after heavy i/o, system stalls with many processes stuck in tstile state

[mlelstv%serpens.de@localhost (Michael van Elst)]

The following reply was made to PR kern/57198; it has been noted by GNATS.

From: mlelstv%serpens.de@localhost (Michael van Elst)
To: gnats-bugs%netbsd.org@localhost
Cc: 
Subject: Re: kern/57198: after heavy i/o, system stalls with many processes stuck in tstile state
Date: Wed, 25 Jan 2023 06:36:39 -0000 (UTC)

 fission%ldx.ca@localhost writes:
 
 >biowait(1042e04e0, 1042e0600, 1b13c00, 1b13c00, fffffffffffffffe, 1042e05d0) at netbsd:biowait+0x34
 >wapbl_buffered_flush(104141600, 1, 0, 1042e05a8, 0, 1042e04e0) at netbsd:wapbl_buffered_flush+0x168
 >wapbl_write_commit(104141600, 37ea00, 1cc000, 1c61c00, 0, 1042ea8c0) at netbsd:wapbl_write_commit+0xc0
 >wapbl_flush(104141600, 0, 104141678, 104141670, 1, 37ea00) at netbsd:wapbl_flush+0x710
 >ffs_sync(103f2d000, 2, 1033e8040, 0, 0, 0) at netbsd:ffs_sync+0x174
 >VFS_SYNC(103f2d000, 3, 1033e8040, 1c7c000, 1, 103f2d000) at netbsd:VFS_SYNC+0x24
 >sched_sync(1ccc018, e0048000, 1c61eec, 1033e41b0, 1ccc020, 0) at netbsd:sched_sync+0x120
 >lwp_trampoline(ffd0d92c, 1140d0, 1173b8, 1, ffe91c98, 0) at netbsd:lwp_trampoline+0x8
 
 
 This means your disk device isn't finishing an I/O request.
 
 All the tstile'd processes are just victims of this.
 
 
 
 >[   1.0000000] cmdide0 at pci1 dev 3 function 0: CMD Technology PCI0646 (rev. 0x07)
 
 I'd suspect a bug in the cmdide driver, or the emulation of that hardware.