> I have attached a very small patch (test9) to remove the kernel lock from
> kernel/acct.c. If I am missing something major (a brain?), I apologise in
> advance. I have tested this on my UP x86 with spinlock debugging. I would
> appreciate comments or an explanation of why this can't be done if you
have
> the time. Thanks.
>
> Nick
>

Maybe there was a possibility of a race when acct_auto_close calls sys_acct
in the last patch. If so, this should fix it.

Nick.

bsdacct2.patch

Reply via email to