* /proc/sys/kernel/l2cr and the 750FX
@ 2003-09-25 16:20 Guy M. Streeter
2003-09-25 18:21 ` Tom Rini
0 siblings, 1 reply; 8+ messages in thread
From: Guy M. Streeter @ 2003-09-25 16:20 UTC (permalink / raw)
To: linuxppc-embedded
[-- Attachment #1: Type: text/plain, Size: 193 bytes --]
The 750FX has a different L2CR register layout than other processors.
The attached patch against 2.4.22 makes /proc/sys/kernel/l2cr show a
better decode of the register for the 750FX.
--Guy
[-- Attachment #2: l2cr750fx.diff --]
[-- Type: text/plain, Size: 3066 bytes --]
--- tmp/arch/ppc/kernel/ppc_htab.c 2003-07-25 09:11:21.000000000 -0500
+++ ppmc750fx/arch/ppc/kernel/ppc_htab.c 2003-09-24 11:21:47.000000000 -0500
@@ -503,40 +503,59 @@ int proc_dol2crvec(ctl_table *table, int
left -= len;
_set_L2CR(val);
} else {
+ int is750fx = PVR_VER(mfspr(PVR)) == PVR_VER(PVR_750FX);
p = buf;
if (!first)
*p++ = '\t';
val = _get_L2CR();
p += sprintf(p, "0x%08x: ", val);
- p += sprintf(p, " L2 %s", (val >> 31) & 1 ? "enabled" :
+ p += sprintf(p, " L2 %s, ", (val >> 31) & 1 ? "enabled" :
"disabled");
- p += sprintf(p, ", %sparity", (val>>30)&1 ? "" : "no ");
+ if (!(val>>30&1))
+ p += sprintf(p, "no ");
+ if (is750fx)
+ p += sprintf(p, "ECC checkstop");
+ else
+ p += sprintf(p, "parity");
/* 75x & 74x0 have different L2CR than 745x */
if (!(cur_cpu_spec[0]->cpu_features &
CPU_FTR_SPEC7450)) {
- p += sprintf(p, ", %s",
- sizestrings[(val >> 28) & 3]);
- p += sprintf(p, ", %s",
- clockstrings[(val >> 25) & 7]);
- p += sprintf(p, ", %s",
- typestrings[(val >> 23) & 3]);
+ if (!is750fx) {
+ p += sprintf(p, ", %s",
+ sizestrings[(val >> 28) & 3]);
+ p += sprintf(p, ", %s",
+ clockstrings[(val >> 25) & 7]);
+ p += sprintf(p, ", %s",
+ typestrings[(val >> 23) & 3]);
+ }
p += sprintf(p, "%s", (val>>22)&1 ?
- ", data only" : "");
- p += sprintf(p, "%s", (val>>20)&1 ?
- ", ZZ enabled": "");
+ ", data only" : "");
+ if (!is750fx) {
+ p += sprintf(p, "%s", (val>>20)&1 ?
+ ", ZZ enabled": "");
+ }
p += sprintf(p, ", %s", (val>>19)&1 ?
"write-through" : "copy-back");
p += sprintf(p, "%s", (val>>18)&1 ?
", testing" : "");
- p += sprintf(p, ", %sns hold",
- holdstrings[(val>>16)&3]);
- p += sprintf(p, "%s", (val>>15)&1 ?
- ", DLL slow" : "");
- p += sprintf(p, "%s", (val>>14)&1 ?
- ", diff clock" :"");
- p += sprintf(p, "%s", (val>>13)&1 ?
- ", DLL bypass" :"");
+ if (!is750fx) {
+ p += sprintf(p, ", %sns hold",
+ holdstrings[(val>>16)&3]);
+ p += sprintf(p, "%s", (val>>15)&1 ?
+ ", DLL slow" : "");
+ p += sprintf(p, "%s", (val>>14)&1 ?
+ ", diff clock" :"");
+ p += sprintf(p, "%s", (val>>13)&1 ?
+ ", DLL bypass" :"");
+ } else {
+ if ((val>>11)&1)
+ p += sprintf(p, ", lock way 0");
+ if ((val>>10)&1)
+ p += sprintf(p, ", lock way 1");
+ if ((val>>9)&1)
+ p += sprintf(p, ", Snoop Hit in Locked Line Error Enabled");
+ }
} else { /* 745x */
p += sprintf(p, ", %sinstn only", (val>>20)&1 ?
"" : "no ");
--- tmp/include/asm-ppc/processor.h 2003-07-25 09:11:30.000000000 -0500
+++ ppmc750fx/include/asm-ppc/processor.h 2003-09-25 11:14:59.000000000 -0500
@@ -685,6 +685,7 @@
#define PVR_750 PVR_740
#define PVR_740P 0x10080000
#define PVR_750P PVR_740P
+#define PVR_750FX 0x70000000
#define PVR_7400 0x000C0000
#define PVR_7410 0x800C0000
#define PVR_7450 0x80000000
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: /proc/sys/kernel/l2cr and the 750FX
2003-09-25 16:20 /proc/sys/kernel/l2cr and the 750FX Guy M. Streeter
@ 2003-09-25 18:21 ` Tom Rini
2003-09-25 19:37 ` Guy M. Streeter
0 siblings, 1 reply; 8+ messages in thread
From: Tom Rini @ 2003-09-25 18:21 UTC (permalink / raw)
To: Guy M. Streeter; +Cc: linuxppc-embedded
On Thu, Sep 25, 2003 at 11:20:39AM -0500, Guy M. Streeter wrote:
> The 750FX has a different L2CR register layout than other processors.
> The attached patch against 2.4.22 makes /proc/sys/kernel/l2cr show a
> better decode of the register for the 750FX.
This shouldn't be needed in 2.4.23-current, can you verify? Also,
adding more PVR_XXXX defines is deprecated, as there are cpufeatures for
such things.
--
Tom Rini
http://gate.crashing.org/~trini/
** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: /proc/sys/kernel/l2cr and the 750FX
2003-09-25 18:21 ` Tom Rini
@ 2003-09-25 19:37 ` Guy M. Streeter
2003-09-25 23:40 ` Tom Rini
0 siblings, 1 reply; 8+ messages in thread
From: Guy M. Streeter @ 2003-09-25 19:37 UTC (permalink / raw)
To: Tom Rini; +Cc: linuxppc-embedded
I don't see any difference in this section of code, in v2.4.23-pre5
Was a change made someplace else to support the 750FX?
--Guy
On Thu, 2003-09-25 at 13:21, Tom Rini wrote:
> On Thu, Sep 25, 2003 at 11:20:39AM -0500, Guy M. Streeter wrote:
>
> > The 750FX has a different L2CR register layout than other processors.
> > The attached patch against 2.4.22 makes /proc/sys/kernel/l2cr show a
> > better decode of the register for the 750FX.
>
> This shouldn't be needed in 2.4.23-current, can you verify? Also,
> adding more PVR_XXXX defines is deprecated, as there are cpufeatures for
> such things.
>
> --
> Tom Rini
> http://gate.crashing.org/~trini/
>
** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: /proc/sys/kernel/l2cr and the 750FX
2003-09-25 19:37 ` Guy M. Streeter
@ 2003-09-25 23:40 ` Tom Rini
0 siblings, 0 replies; 8+ messages in thread
From: Tom Rini @ 2003-09-25 23:40 UTC (permalink / raw)
To: Guy M. Streeter; +Cc: linuxppc-embedded
On Thu, Sep 25, 2003 at 02:37:56PM -0500, Guy M. Streeter wrote:
> I don't see any difference in this section of code, in v2.4.23-pre5
> Was a change made someplace else to support the 750FX?
Support was added for a number of newer CPUs, iirc. I don't recall of
the 750FX was one of them.
--
Tom Rini
http://gate.crashing.org/~trini/
** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/
^ permalink raw reply [flat|nested] 8+ messages in thread
* /proc/sys/kernel/l2cr and the 750FX
@ 2003-11-14 15:44 Guy M. Streeter
2003-11-14 16:27 ` Tom Rini
0 siblings, 1 reply; 8+ messages in thread
From: Guy M. Streeter @ 2003-11-14 15:44 UTC (permalink / raw)
To: linuxppc-embedded
[-- Attachment #1: Type: text/plain, Size: 548 bytes --]
In September I posted a version of this patch, which corrects the
formatted display of the L2CR register for 750FX processors.
Tom Rini suggested that it should use a cpu feature flag instead of the
processors version register, and also that the change might already have
been made in later kernel source.
I have check all the 2.4.23+ and 2.6.0 source trees, and I don't see
any fix for this. I have changed the patch to use a feature flag instead
of checking the PVR. I would like to see this change considered for
inclusion.
thanks,
--Guy
[-- Attachment #2: l2cr750fx.diff --]
[-- Type: text/plain, Size: 4544 bytes --]
--- v2.4.22_linuxppc_2_4_devel-ac4-lolat/arch/ppc/kernel/cputable.c 2003-07-25 14:39:48.000000000 +0000
+++ ppmc750fx/arch/ppc/kernel/cputable.c 2003-10-27 22:40:46.000000000 +0000
@@ -167,7 +167,7 @@ struct cpu_spec cpu_specs[] = {
0xffffff00, 0x70000100, "750FX",
CPU_FTR_SPLIT_ID_CACHE | CPU_FTR_CAN_DOZE | CPU_FTR_USE_TB |
CPU_FTR_L2CR | CPU_FTR_TAU | CPU_FTR_HPTE_TABLE | CPU_FTR_CAN_NAP |
- CPU_FTR_DUAL_PLL_750FX | CPU_FTR_NO_DPM,
+ CPU_FTR_DUAL_PLL_750FX | CPU_FTR_NO_DPM | CPU_FTR_750FX_L2CR,
COMMON_PPC,
32, 32,
__setup_cpu_750
@@ -176,7 +176,8 @@ struct cpu_spec cpu_specs[] = {
0xffffffff, 0x70000200, "750FX",
CPU_FTR_SPLIT_ID_CACHE | CPU_FTR_CAN_DOZE | CPU_FTR_USE_TB |
CPU_FTR_L2CR | CPU_FTR_TAU | CPU_FTR_HPTE_TABLE | CPU_FTR_CAN_NAP |
- CPU_FTR_DUAL_PLL_750FX | CPU_FTR_HAS_HIGH_BATS | CPU_FTR_NO_DPM,
+ CPU_FTR_DUAL_PLL_750FX | CPU_FTR_HAS_HIGH_BATS | CPU_FTR_NO_DPM |
+ CPU_FTR_750FX_L2CR,
COMMON_PPC,
32, 32,
__setup_cpu_750
@@ -185,7 +186,7 @@ struct cpu_spec cpu_specs[] = {
0xffff0000, 0x70000000, "750FX",
CPU_FTR_SPLIT_ID_CACHE | CPU_FTR_CAN_DOZE | CPU_FTR_USE_TB |
CPU_FTR_L2CR | CPU_FTR_TAU | CPU_FTR_HPTE_TABLE | CPU_FTR_CAN_NAP |
- CPU_FTR_DUAL_PLL_750FX | CPU_FTR_HAS_HIGH_BATS,
+ CPU_FTR_DUAL_PLL_750FX | CPU_FTR_HAS_HIGH_BATS | CPU_FTR_750FX_L2CR,
COMMON_PPC,
32, 32,
__setup_cpu_750fx
--- v2.4.22_linuxppc_2_4_devel-ac4-lolat/arch/ppc/kernel/ppc_htab.c 2003-07-25 14:11:21.000000000 +0000
+++ ppmc750fx/arch/ppc/kernel/ppc_htab.c 2003-10-27 22:44:02.000000000 +0000
@@ -503,40 +503,59 @@ int proc_dol2crvec(ctl_table *table, int
left -= len;
_set_L2CR(val);
} else {
+ int is750fx = cur_cpu_spec[0]->cpu_features & CPU_FTR_750FX_L2CR;
p = buf;
if (!first)
*p++ = '\t';
val = _get_L2CR();
p += sprintf(p, "0x%08x: ", val);
- p += sprintf(p, " L2 %s", (val >> 31) & 1 ? "enabled" :
+ p += sprintf(p, " L2 %s, ", (val >> 31) & 1 ? "enabled" :
"disabled");
- p += sprintf(p, ", %sparity", (val>>30)&1 ? "" : "no ");
+ if (!(val>>30&1))
+ p += sprintf(p, "no ");
+ if (is750fx)
+ p += sprintf(p, "ECC checkstop");
+ else
+ p += sprintf(p, "parity");
/* 75x & 74x0 have different L2CR than 745x */
if (!(cur_cpu_spec[0]->cpu_features &
CPU_FTR_SPEC7450)) {
- p += sprintf(p, ", %s",
- sizestrings[(val >> 28) & 3]);
- p += sprintf(p, ", %s",
- clockstrings[(val >> 25) & 7]);
- p += sprintf(p, ", %s",
- typestrings[(val >> 23) & 3]);
+ if (!is750fx) {
+ p += sprintf(p, ", %s",
+ sizestrings[(val >> 28) & 3]);
+ p += sprintf(p, ", %s",
+ clockstrings[(val >> 25) & 7]);
+ p += sprintf(p, ", %s",
+ typestrings[(val >> 23) & 3]);
+ }
p += sprintf(p, "%s", (val>>22)&1 ?
- ", data only" : "");
- p += sprintf(p, "%s", (val>>20)&1 ?
- ", ZZ enabled": "");
+ ", data only" : "");
+ if (!is750fx) {
+ p += sprintf(p, "%s", (val>>20)&1 ?
+ ", ZZ enabled": "");
+ }
p += sprintf(p, ", %s", (val>>19)&1 ?
"write-through" : "copy-back");
p += sprintf(p, "%s", (val>>18)&1 ?
", testing" : "");
- p += sprintf(p, ", %sns hold",
- holdstrings[(val>>16)&3]);
- p += sprintf(p, "%s", (val>>15)&1 ?
- ", DLL slow" : "");
- p += sprintf(p, "%s", (val>>14)&1 ?
- ", diff clock" :"");
- p += sprintf(p, "%s", (val>>13)&1 ?
- ", DLL bypass" :"");
+ if (!is750fx) {
+ p += sprintf(p, ", %sns hold",
+ holdstrings[(val>>16)&3]);
+ p += sprintf(p, "%s", (val>>15)&1 ?
+ ", DLL slow" : "");
+ p += sprintf(p, "%s", (val>>14)&1 ?
+ ", diff clock" :"");
+ p += sprintf(p, "%s", (val>>13)&1 ?
+ ", DLL bypass" :"");
+ } else {
+ if ((val>>11)&1)
+ p += sprintf(p, ", lock way 0");
+ if ((val>>10)&1)
+ p += sprintf(p, ", lock way 1");
+ if ((val>>9)&1)
+ p += sprintf(p, ", Snoop Hit in Locked Line Error Enabled");
+ }
} else { /* 745x */
p += sprintf(p, ", %sinstn only", (val>>20)&1 ?
"" : "no ");
--- v2.4.22_linuxppc_2_4_devel-ac4-lolat/include/asm-ppc/cputable.h 2003-05-10 10:43:37.000000000 +0000
+++ ppmc750fx/include/asm-ppc/cputable.h 2003-10-27 22:40:38.000000000 +0000
@@ -75,6 +75,7 @@ extern struct cpu_spec *cur_cpu_spec[];
#define CPU_FTR_DUAL_PLL_750FX 0x00004000
#define CPU_FTR_NO_DPM 0x00008000
#define CPU_FTR_HAS_HIGH_BATS 0x00010000
+#define CPU_FTR_750FX_L2CR 0x00020000
#ifdef __ASSEMBLY__
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: /proc/sys/kernel/l2cr and the 750FX
2003-11-14 15:44 Guy M. Streeter
@ 2003-11-14 16:27 ` Tom Rini
2003-11-14 17:09 ` Tom Rini
0 siblings, 1 reply; 8+ messages in thread
From: Tom Rini @ 2003-11-14 16:27 UTC (permalink / raw)
To: Guy M. Streeter; +Cc: linuxppc-embedded
On Fri, Nov 14, 2003 at 09:44:01AM -0600, Guy M. Streeter wrote:
> In September I posted a version of this patch, which corrects the
> formatted display of the L2CR register for 750FX processors.
> Tom Rini suggested that it should use a cpu feature flag instead of the
> processors version register, and also that the change might already have
> been made in later kernel source.
> I have check all the 2.4.23+ and 2.6.0 source trees, and I don't see
> any fix for this. I have changed the patch to use a feature flag instead
> of checking the PVR. I would like to see this change considered for
> inclusion.
I've checked this into my 2.4.24-pre tree.
--
Tom Rini
http://gate.crashing.org/~trini/
** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: /proc/sys/kernel/l2cr and the 750FX
2003-11-14 16:27 ` Tom Rini
@ 2003-11-14 17:09 ` Tom Rini
2003-11-14 17:22 ` Guy M. Streeter
0 siblings, 1 reply; 8+ messages in thread
From: Tom Rini @ 2003-11-14 17:09 UTC (permalink / raw)
To: Guy M. Streeter; +Cc: linuxppc-embedded
On Fri, Nov 14, 2003 at 09:27:45AM -0700, Tom Rini wrote:
>
> On Fri, Nov 14, 2003 at 09:44:01AM -0600, Guy M. Streeter wrote:
>
> > In September I posted a version of this patch, which corrects the
> > formatted display of the L2CR register for 750FX processors.
> > Tom Rini suggested that it should use a cpu feature flag instead of the
> > processors version register, and also that the change might already have
> > been made in later kernel source.
> > I have check all the 2.4.23+ and 2.6.0 source trees, and I don't see
> > any fix for this. I have changed the patch to use a feature flag instead
> > of checking the PVR. I would like to see this change considered for
> > inclusion.
>
> I've checked this into my 2.4.24-pre tree.
.. with a change to turn CPU_FTR_DUAL_PLL_750FX into CPU_FTR_750FX,
since that is applied to all revs of the 750FX.
--
Tom Rini
http://gate.crashing.org/~trini/
** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: /proc/sys/kernel/l2cr and the 750FX
2003-11-14 17:09 ` Tom Rini
@ 2003-11-14 17:22 ` Guy M. Streeter
0 siblings, 0 replies; 8+ messages in thread
From: Guy M. Streeter @ 2003-11-14 17:22 UTC (permalink / raw)
To: Tom Rini; +Cc: linuxppc-embedded
On Fri, 2003-11-14 at 11:09, Tom Rini wrote:
> On Fri, Nov 14, 2003 at 09:27:45AM -0700, Tom Rini wrote:
> >
> > On Fri, Nov 14, 2003 at 09:44:01AM -0600, Guy M. Streeter wrote:
> >
> > > In September I posted a version of this patch, which corrects the
> > > formatted display of the L2CR register for 750FX processors.
> > > Tom Rini suggested that it should use a cpu feature flag instead of the
> > > processors version register, and also that the change might already have
> > > been made in later kernel source.
> > > I have check all the 2.4.23+ and 2.6.0 source trees, and I don't see
> > > any fix for this. I have changed the patch to use a feature flag instead
> > > of checking the PVR. I would like to see this change considered for
> > > inclusion.
> >
> > I've checked this into my 2.4.24-pre tree.
>
> .. with a change to turn CPU_FTR_DUAL_PLL_750FX into CPU_FTR_750FX,
> since that is applied to all revs of the 750FX.
Cool. I had thought about doing that, and then my mind wandered off. :)
--Guy
** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/
^ permalink raw reply [flat|nested] 8+ messages in thread
end of thread, other threads:[~2003-11-14 17:22 UTC | newest]
Thread overview: 8+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2003-09-25 16:20 /proc/sys/kernel/l2cr and the 750FX Guy M. Streeter
2003-09-25 18:21 ` Tom Rini
2003-09-25 19:37 ` Guy M. Streeter
2003-09-25 23:40 ` Tom Rini
-- strict thread matches above, loose matches on Subject: below --
2003-11-14 15:44 Guy M. Streeter
2003-11-14 16:27 ` Tom Rini
2003-11-14 17:09 ` Tom Rini
2003-11-14 17:22 ` Guy M. Streeter
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).