On Wed, 11 Jun 2008, James Carlson wrote:
> [EMAIL PROTECTED] writes:
>> to that driver in order to perform yet another guards check; Should
>> kmem_alloc() be fixed ?
>
> Seems like an obvious case to me; yes.
Hmm, I've found:
http://bugs.opensolaris.org/view_bug.do?bug_id=4241150
which created the "size == 0" panic from where before all hell would break
loose ...
and since then (1999), most such panics, if they occurred, have been
caused by drivers passing e.g. uninitialized / unvalidated values into
kmem_alloc() - and were fixed by fixing the driver bug.
I've got to ping a few people about that. In the case of the code I'm on
right now, the behaviour "try KM_NOSLEEP on huge buffers" is deliberate,
and any limit not imposed by <system resources available> would break
driver functionality.
I do think the abovementioned codechange, which introduced:
1049 if (size == 0)
1050 panic("vmem_xalloc(): size == 0");
in vmem_xalloc(), needs to be modified like this:
if (size == 0)
if (vmflag & VM_NOSLEEP)
return (NULL);
else
panic("vmem_xalloc(): size == 0");
to allow KM_NOSLEEP users graceful recovery ?
FrankH.
_______________________________________________
opensolaris-code mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/opensolaris-code
