> It's not just that, if you always have to cover your behind when
> doing tsleep you may wind up masking wakeup bugs. Places like
> "vfs_bio.c" line 586 of 3182:
>
> bp->b_xflags |= BX_BKGRDWAIT;
> tsleep(&bp->b_xflags, PRIBIO, "biord", 0);
> if (bp->b_xflags & BX_BKGRDINPROG)
> panic("bwrite: still writing");
> }
>
> If replaced by a while() _may_ obscure a buffercache bug.
>
> Personally I'd like to be able to catch such bugs than let them go
> because the API (wakeups can happen at any time) prohibits this.
No in a fully threaded world.....
Nate
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message