Commit 1292500b replaced
"=m" (*field) : "1" (*field)
with
"=m" (*field) :
with comment "The following patch fixes it by using the '+' operator on
the (*field) operand, marking it as read-write to gcc."
'+' was actually forgotten. This really puts it.
Signed-off-by: Samuel Thibault <[email protected]>
diff --git a/drivers/scsi/ultrastor.c b/drivers/scsi/ultrastor.c
index 0571ef9..dc076e0 100644
--- a/drivers/scsi/ultrastor.c
+++ b/drivers/scsi/ultrastor.c
@@ -306,7 +306,7 @@ static inline int find_and_clear_bit_16(unsigned long *field)
"0: bsfw %1,%w0\n\t"
"btr %0,%1\n\t"
"jnc 0b"
- : "=&r" (rv), "=m" (*field) :);
+ : "=&r" (rv), "+m" (*field) :);
return rv;
}
On Wed, May 18, 2011 at 05:06:05PM +0200, Samuel Thibault wrote:
> Commit 1292500b replaced
>
> "=m" (*field) : "1" (*field)
>
> with
>
> "=m" (*field) :
>
> with comment "The following patch fixes it by using the '+' operator on
> the (*field) operand, marking it as read-write to gcc."
> '+' was actually forgotten. This really puts it.
Do you actually have the hardware or was this just a code audit?
I have the strong suspicion that this driver is pretty much dead and
bitrotting.
Christoph Hellwig, le Fri 20 May 2011 09:02:15 -0400, a ?crit :
> On Wed, May 18, 2011 at 05:06:05PM +0200, Samuel Thibault wrote:
> > Commit 1292500b replaced
> >
> > "=m" (*field) : "1" (*field)
> >
> > with
> >
> > "=m" (*field) :
> >
> > with comment "The following patch fixes it by using the '+' operator on
> > the (*field) operand, marking it as read-write to gcc."
> > '+' was actually forgotten. This really puts it.
>
> Do you actually have the hardware or was this just a code audit?
Just a code audit.
> I have the strong suspicion that this driver is pretty much dead and
> bitrotting.
Most probably, yes.
Samuel
On Fri, 20 May 2011 09:02:15 -0400
Christoph Hellwig <[email protected]> wrote:
> On Wed, May 18, 2011 at 05:06:05PM +0200, Samuel Thibault wrote:
> > Commit 1292500b replaced
> >
> > "=m" (*field) : "1" (*field)
> >
> > with
> >
> > "=m" (*field) :
> >
> > with comment "The following patch fixes it by using the '+' operator on
> > the (*field) operand, marking it as read-write to gcc."
> > '+' was actually forgotten. This really puts it.
>
> Do you actually have the hardware or was this just a code audit?
>
> I have the strong suspicion that this driver is pretty much dead and
> bitrotting.
There are a pile of scsi drivers like the Ultrastor that probably want
retiring via staging.
I can believe an AHA152x or two lurk around, and the odd NCR53c80 and
clones because they were so voluminous but the rest I doubt somewhat.
On Fri, 2011-05-20 at 14:24 +0100, Alan Cox wrote:
> On Fri, 20 May 2011 09:02:15 -0400
> Christoph Hellwig <[email protected]> wrote:
>
> > On Wed, May 18, 2011 at 05:06:05PM +0200, Samuel Thibault wrote:
> > > Commit 1292500b replaced
> > >
> > > "=m" (*field) : "1" (*field)
> > >
> > > with
> > >
> > > "=m" (*field) :
> > >
> > > with comment "The following patch fixes it by using the '+' operator on
> > > the (*field) operand, marking it as read-write to gcc."
> > > '+' was actually forgotten. This really puts it.
> >
> > Do you actually have the hardware or was this just a code audit?
> >
> > I have the strong suspicion that this driver is pretty much dead and
> > bitrotting.
>
> There are a pile of scsi drivers like the Ultrastor that probably want
> retiring via staging.
OK, can someone work out how we do this and then send the patch (I
assume it's just an update in the deprecated features and a move to
staging?)
> I can believe an AHA152x or two lurk around,
Yes, got one ... and occasionally even test with it.
> and the odd NCR53c80 and
> clones because they were so voluminous but the rest I doubt somewhat.
James