[PATCH] fix phys. address conversion in ia64_pal_tr_read

[PATCH] fix phys. address conversion in ia64_pal_tr_read

[Jesse Barnes]

[-- Attachment #1: Type: text/plain, Size: 356 bytes --]

The palinfo module may call ia64_pal_tr_read using an IPI.  If the processor 
receiving the IPI is idle, the call will fail since the idle process stack is 
in region 5 and ia64_pal_tr_read uses __pa rather than ia64_tpa.  This small 
patch fixes the problem by making it use ia64_tpa instead.

Signed-off-by: Jesse Barnes <jbarnes@sgi.com>

Thanks,
Jesse

[-- Attachment #2: tr-info-fix-2.patch --]
[-- Type: text/plain, Size: 551 bytes --]

===== include/asm-ia64/pal.h 1.9 vs edited =====
--- 1.9/include/asm-ia64/pal.h	2003-11-25 00:23:21 -08:00
+++ edited/include/asm-ia64/pal.h	2004-11-15 17:07:51 -08:00
@@ -1531,7 +1531,7 @@
 ia64_pal_tr_read (u64 reg_num, u64 tr_type, u64 *tr_buffer, pal_tr_valid_u_t *tr_valid)
 {
 	struct ia64_pal_retval iprv;
-	PAL_CALL_PHYS_STK(iprv, PAL_VM_TR_READ, reg_num, tr_type,(u64)__pa(tr_buffer));
+	PAL_CALL_PHYS_STK(iprv, PAL_VM_TR_READ, reg_num, tr_type,(u64)ia64_tpa(tr_buffer));
 	if (tr_valid)
 		tr_valid->piv_val = iprv.v0;
 	return iprv.status;

Re: [PATCH] fix phys. address conversion in ia64_pal_tr_read

[David Mosberger]

>>>>> On Mon, 15 Nov 2004 17:22:41 -0800, Jesse Barnes <jbarnes@engr.sgi.com> said:

  Jesse> The palinfo module may call ia64_pal_tr_read using an IPI.
  Jesse> If the processor receiving the IPI is idle, the call will
  Jesse> fail since the idle process stack is in region 5 and
  Jesse> ia64_pal_tr_read uses __pa rather than ia64_tpa.  This small
  Jesse> patch fixes the problem by making it use ia64_tpa instead.

Perhaps its worthwhile to do any audit for this kind of problems?  It
may not be very difficult to hack sparse to add a check for __pa()
calls on (stack-)local variables.

	--david

Re: [PATCH] fix phys. address conversion in ia64_pal_tr_read

[Jesse Barnes]

On Monday, November 15, 2004 6:03 pm, David Mosberger wrote:
> >>>>> On Mon, 15 Nov 2004 17:22:41 -0800, Jesse Barnes
> >>>>> <jbarnes@engr.sgi.com> said:
>
>   Jesse> The palinfo module may call ia64_pal_tr_read using an IPI.
>   Jesse> If the processor receiving the IPI is idle, the call will
>   Jesse> fail since the idle process stack is in region 5 and
>   Jesse> ia64_pal_tr_read uses __pa rather than ia64_tpa.  This small
>   Jesse> patch fixes the problem by making it use ia64_tpa instead.
>
> Perhaps its worthwhile to do any audit for this kind of problems?  It
> may not be very difficult to hack sparse to add a check for __pa()
> calls on (stack-)local variables.

I didn't see anything obvious in a quick check of all the ia64 users of __pa 
(they were mostly referring to stuff in the static data section), but I may 
have missed something.

Jesse

Re: [PATCH] fix phys. address conversion in ia64_pal_tr_read

[David Mosberger]

>>>>> On Tue, 16 Nov 2004 09:01:49 -0800, Jesse Barnes <jbarnes@engr.sgi.com> said:

  Jesse> On Monday, November 15, 2004 6:03 pm, David Mosberger wrote:
  >> >>>>> On Mon, 15 Nov 2004 17:22:41 -0800, Jesse Barnes >>>>>
  >> <jbarnes@engr.sgi.com> said:

  Jesse> The palinfo module may call ia64_pal_tr_read using an IPI.
  Jesse> If the processor receiving the IPI is idle, the call will
  Jesse> fail since the idle process stack is in region 5 and
  Jesse> ia64_pal_tr_read uses __pa rather than ia64_tpa.  This small
  Jesse> patch fixes the problem by making it use ia64_tpa instead.
  >>  Perhaps its worthwhile to do any audit for this kind of
  >> problems?  It may not be very difficult to hack sparse to add a
  >> check for __pa() calls on (stack-)local variables.

  Jesse> I didn't see anything obvious in a quick check of all the
  Jesse> ia64 users of __pa (they were mostly referring to stuff in
  Jesse> the static data section), but I may have missed something.

OK, thanks for looking!

	--david