[BUG] 2.6.24-rc3-git2 softlockup detected

[Andrew Morton]

On Fri, 30 Nov 2007 12:58:06 +0530
Kamalesh Babulal <kamalesh at linux.vnet.ibm.com> wrote:

> Andrew Morton wrote:
> > On Thu, 29 Nov 2007 23:00:47 -0800 Andrew Morton <akpm at linux-foundation.org> wrote:
> > 
> >> On Fri, 30 Nov 2007 01:39:29 -0500 Kyle McMartin <kyle at mcmartin.ca> wrote:
> >>
> >>> On Thu, Nov 29, 2007 at 12:35:33AM -0800, Andrew Morton wrote:
> >>>> ten million is close enough to infinity for me to assume that we broke the
> >>>> driver and that's never going to terminate.
> >>>>
> >>> how about this? doesn't break things on my pa8800:
> >>>
> >>> diff --git a/drivers/scsi/sym53c8xx_2/sym_hipd.c b/drivers/scsi/sym53c8xx_2/sym_hipd.c
> >>> index 463f119..ef01cb1 100644
> >>> --- a/drivers/scsi/sym53c8xx_2/sym_hipd.c
> >>> +++ b/drivers/scsi/sym53c8xx_2/sym_hipd.c
> >>> @@ -1037,10 +1037,13 @@ restart_test:
> >>>  	/*
> >>>  	 *  Wait 'til done (with timeout)
> >>>  	 */
> >>> -	for (i=0; i<SYM_SNOOP_TIMEOUT; i++)
> >>> +	do {	
> >>>  		if (INB(np, nc_istat) & (INTF|SIP|DIP))
> >>>  			break;
> >>> -	if (i>=SYM_SNOOP_TIMEOUT) {
> >>> +		msleep(10);
> >>> +	} while (i++ < SYM_SNOOP_TIMEOUT);
> >>> +
> >>> +	if (i >= SYM_SNOOP_TIMEOUT) {
> >>>  		printf ("CACHE TEST FAILED: timeout.\n");
> >>>  		return (0x20);
> >>>  	}
> >>> diff --git a/drivers/scsi/sym53c8xx_2/sym_hipd.h b/drivers/scsi/sym53c8xx_2/sym_hipd.h
> >>> index ad07880..85c483b 100644
> >>> --- a/drivers/scsi/sym53c8xx_2/sym_hipd.h
> >>> +++ b/drivers/scsi/sym53c8xx_2/sym_hipd.h
> >>> @@ -339,7 +339,7 @@
> >>>  /*
> >>>   *  Misc.
> >>>   */
> >>> -#define SYM_SNOOP_TIMEOUT (10000000)
> >>> +#define SYM_SNOOP_TIMEOUT (1000)
> >>>  #define BUS_8_BIT	0
> >>>  #define BUS_16_BIT	1
> >>>  
> >> That might be the fix, but do we know what we're actually fixing?  afaik
> >> 2.6.24-rc3 doesn't get this timeout, 2.6.24-rc3-mm2 does get it and we
> >> don't know why?
> >>
> > 
> > <looks at Subject:>
> > 
> > <Checks that Rafael was cc'ed>
> > 
> > So 2.6.24-rc3 was OK and 2.6.24-rc3-git2 is not?
> 
> Yes, the 2.6.24-rc3 was Ok and this is seen from 2.6.24-rc3-git2/3/4.
> 

There are effectively no drivers/scsi/ changes after 2.6.24-rc3 and we
don't (I believe) have a clue what caused this regression.

Can you please do a bisection search on this?

Thanks.