Re: kern/55237

[Andrew Doran <ad%netbsd.org@localhost>]

On Sat, May 23, 2020 at 09:50:02PM +0000, Yorick Hardy wrote:
> The following reply was made to PR kern/55237; it has been noted by GNATS.
> 
> From: Yorick Hardy <yorickhardy%gmail.com@localhost>
> To: gnats-bugs%netbsd.org@localhost
> Cc: 
> Subject: Re: kern/55237
> Date: Sat, 23 May 2020 23:47:11 +0200
> 
>  A very wild guess, I was wondering if this change affects the writecount
>  (after a superficial reading of fd_putfile):
>  
>   cvs rdiff -u -r1.543 -r1.544 src/sys/kern/vfs_syscalls.c
>  
>  mail/fdm depends on databases/tdb, which uses fdatasync (checked only by a quick grep).
>  
>  I am building a test kernel with the change reverted, and will try it out for
>  a few days to see if it is stable.

At a guess is not responsible, but worthwhile trying with it reverted.

I made a commit but I don't think that one will cure it either.

I'm suspicious of the assertion actually, there might be a race condition. 
Will think about it some more.

Andrew