* Re: [patch 2.4 1/2] backport 2.6 x86 cpu capabilities
@ 2003-08-08 19:41 Mikael Pettersson
2003-08-09 14:05 ` Jeff Garzik
0 siblings, 1 reply; 8+ messages in thread
From: Mikael Pettersson @ 2003-08-08 19:41 UTC (permalink / raw)
To: jgarzik; +Cc: linux-kernel
On Fri, 08 Aug 2003 09:15:03 -0400, Jeff Garzik wrote:
>> If you change NCAPINTS you also have to change the hardcoded
>> struct offset X86_VENDOR_ID in arch/i386/kernel/head.S. Otherwise
>> nasty stuff happen at boot since boot_cpu_data gets broken.
>
>
>hmmm, reality doesn't seem to bear that out... I made the same change
>to 2.6, without touching head.S, and life continues without "nasty
>stuff" AFAICS.
>
>Do both 2.4 and 2.6 need this change? And, why didn't 2.6 break?
2.4.21-rc1 with NCAPINTS==6 hangs at boot in the local
APIC timer calibration step; before that it detected a
0MHz bus clock and the local APIC NMI watchdog was stuck.
Correcting head.S:X86_VENDOR_ID fixes these problems.
Without correcting head.S:X86_VENDOR_ID, head.S will store
the vendor id partly in the capabilities array. This breaks
both the capabilities and the vendor id. I can't say why 2.6
works, but obviously the CPU setup code has changed since 2.4.
BTW, the patch below should be applied to 2.6.
/Mikael
--- linux-2.6.0-test2/arch/i386/kernel/head.S.~1~ 2003-05-28 22:15:58.000000000 +0200
+++ linux-2.6.0-test2/arch/i386/kernel/head.S 2003-08-08 21:12:42.000000000 +0200
@@ -35,7 +35,7 @@
#define X86_HARD_MATH CPU_PARAMS+6
#define X86_CPUID CPU_PARAMS+8
#define X86_CAPABILITY CPU_PARAMS+12
-#define X86_VENDOR_ID CPU_PARAMS+28
+#define X86_VENDOR_ID CPU_PARAMS+36
/*
* Initialize page tables
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [patch 2.4 1/2] backport 2.6 x86 cpu capabilities
2003-08-08 19:41 [patch 2.4 1/2] backport 2.6 x86 cpu capabilities Mikael Pettersson
@ 2003-08-09 14:05 ` Jeff Garzik
0 siblings, 0 replies; 8+ messages in thread
From: Jeff Garzik @ 2003-08-09 14:05 UTC (permalink / raw)
To: Mikael Pettersson; +Cc: linux-kernel
Mikael Pettersson wrote:
> On Fri, 08 Aug 2003 09:15:03 -0400, Jeff Garzik wrote:
>
>>>If you change NCAPINTS you also have to change the hardcoded
>>>struct offset X86_VENDOR_ID in arch/i386/kernel/head.S. Otherwise
>>>nasty stuff happen at boot since boot_cpu_data gets broken.
>>
>>
>>hmmm, reality doesn't seem to bear that out... I made the same change
>>to 2.6, without touching head.S, and life continues without "nasty
>>stuff" AFAICS.
>>
>>Do both 2.4 and 2.6 need this change? And, why didn't 2.6 break?
>
>
> 2.4.21-rc1 with NCAPINTS==6 hangs at boot in the local
> APIC timer calibration step; before that it detected a
> 0MHz bus clock and the local APIC NMI watchdog was stuck.
> Correcting head.S:X86_VENDOR_ID fixes these problems.
>
> Without correcting head.S:X86_VENDOR_ID, head.S will store
> the vendor id partly in the capabilities array. This breaks
> both the capabilities and the vendor id. I can't say why 2.6
> works, but obviously the CPU setup code has changed since 2.4.
>
> BTW, the patch below should be applied to 2.6.
Thanks for the patch, and for explaining.
Jeff
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [patch 2.4 1/2] backport 2.6 x86 cpu capabilities
2003-08-10 23:59 Albert Cahalan
@ 2003-08-11 0:28 ` Jeff Garzik
0 siblings, 0 replies; 8+ messages in thread
From: Jeff Garzik @ 2003-08-11 0:28 UTC (permalink / raw)
To: Albert Cahalan; +Cc: linux-kernel mailing list, mikpe, m.c.p
Albert Cahalan wrote:
> Mikael Pettersson writes:
>
>
>>2.4.21-rc1 with NCAPINTS==6 hangs at boot in the local
>>APIC timer calibration step; before that it detected a
>>0MHz bus clock and the local APIC NMI watchdog was stuck.
>>Correcting head.S:X86_VENDOR_ID fixes these problems.
>>
>>Without correcting head.S:X86_VENDOR_ID, head.S will store
>>the vendor id partly in the capabilities array. This breaks
>>both the capabilities and the vendor id. I can't say why 2.6
>>works, but obviously the CPU setup code has changed since 2.4.
>
>
> I may be stating the obvious, but in case not...
>
> If Jeff Garzik missed this, others will too. I hope
> that a big comment gets added in both places, assuming
> that automatic offset generation isn't practical.
Yeah, I'm queueing a change to do that, actually, so that a grep for
NCAPINTS will hit head.S.
Jeff
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [patch 2.4 1/2] backport 2.6 x86 cpu capabilities
@ 2003-08-10 23:59 Albert Cahalan
2003-08-11 0:28 ` Jeff Garzik
0 siblings, 1 reply; 8+ messages in thread
From: Albert Cahalan @ 2003-08-10 23:59 UTC (permalink / raw)
To: linux-kernel mailing list; +Cc: jgarzik, mikpe, m.c.p
Mikael Pettersson writes:
> 2.4.21-rc1 with NCAPINTS==6 hangs at boot in the local
> APIC timer calibration step; before that it detected a
> 0MHz bus clock and the local APIC NMI watchdog was stuck.
> Correcting head.S:X86_VENDOR_ID fixes these problems.
>
> Without correcting head.S:X86_VENDOR_ID, head.S will store
> the vendor id partly in the capabilities array. This breaks
> both the capabilities and the vendor id. I can't say why 2.6
> works, but obviously the CPU setup code has changed since 2.4.
I may be stating the obvious, but in case not...
If Jeff Garzik missed this, others will too. I hope
that a big comment gets added in both places, assuming
that automatic offset generation isn't practical.
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [patch 2.4 1/2] backport 2.6 x86 cpu capabilities
2003-08-08 13:15 ` Jeff Garzik
@ 2003-08-09 9:51 ` Marc-Christian Petersen
0 siblings, 0 replies; 8+ messages in thread
From: Marc-Christian Petersen @ 2003-08-09 9:51 UTC (permalink / raw)
To: Jeff Garzik, Mikael Pettersson; +Cc: linux-kernel
On Friday 08 August 2003 15:15, Jeff Garzik wrote:
Hi Jeff,
> >>-#define NCAPINTS 4 /* Currently we have 4 32-bit words worth of info */
> >>+#define NCAPINTS 6 /* Currently we have 6 32-bit words worth of info */
> > If you change NCAPINTS you also have to change the hardcoded
> > struct offset X86_VENDOR_ID in arch/i386/kernel/head.S. Otherwise
> > nasty stuff happen at boot since boot_cpu_data gets broken.
> hmmm, reality doesn't seem to bear that out... I made the same change
> to 2.6, without touching head.S, and life continues without "nasty
> stuff" AFAICS.
> Do both 2.4 and 2.6 need this change? And, why didn't 2.6 break?
Mikael is right. At least 2.4 need this change, otherwise APIC may break.
2.6 might break also in some cases.
ciao, Marc
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [patch 2.4 1/2] backport 2.6 x86 cpu capabilities
2003-08-08 11:19 Mikael Pettersson
@ 2003-08-08 13:15 ` Jeff Garzik
2003-08-09 9:51 ` Marc-Christian Petersen
0 siblings, 1 reply; 8+ messages in thread
From: Jeff Garzik @ 2003-08-08 13:15 UTC (permalink / raw)
To: Mikael Pettersson; +Cc: linux-kernel
Mikael Pettersson wrote:
> On Thu, 7 Aug 2003 22:54:30 -0400, Jeff Garzik wrote:
>
>>(hopefully destined for 2.4.23-pre1)
>>
>>#
>># include/asm-i386/msr.h 1.8 -> 1.9
>># include/asm-i386/cpufeature.h 1.5 -> 1.6
>># arch/i386/kernel/setup.c 1.70 -> 1.71
>>#
>
> ...
>
>>-#define NCAPINTS 4 /* Currently we have 4 32-bit words worth of info */
>>+#define NCAPINTS 6 /* Currently we have 6 32-bit words worth of info */
>
>
> If you change NCAPINTS you also have to change the hardcoded
> struct offset X86_VENDOR_ID in arch/i386/kernel/head.S. Otherwise
> nasty stuff happen at boot since boot_cpu_data gets broken.
hmmm, reality doesn't seem to bear that out... I made the same change
to 2.6, without touching head.S, and life continues without "nasty
stuff" AFAICS.
Do both 2.4 and 2.6 need this change? And, why didn't 2.6 break?
Jeff
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [patch 2.4 1/2] backport 2.6 x86 cpu capabilities
@ 2003-08-08 11:19 Mikael Pettersson
2003-08-08 13:15 ` Jeff Garzik
0 siblings, 1 reply; 8+ messages in thread
From: Mikael Pettersson @ 2003-08-08 11:19 UTC (permalink / raw)
To: jgarzik, linux-kernel
On Thu, 7 Aug 2003 22:54:30 -0400, Jeff Garzik wrote:
>(hopefully destined for 2.4.23-pre1)
>
>#
># include/asm-i386/msr.h 1.8 -> 1.9
># include/asm-i386/cpufeature.h 1.5 -> 1.6
># arch/i386/kernel/setup.c 1.70 -> 1.71
>#
...
>-#define NCAPINTS 4 /* Currently we have 4 32-bit words worth of info */
>+#define NCAPINTS 6 /* Currently we have 6 32-bit words worth of info */
If you change NCAPINTS you also have to change the hardcoded
struct offset X86_VENDOR_ID in arch/i386/kernel/head.S. Otherwise
nasty stuff happen at boot since boot_cpu_data gets broken.
/Mikael
^ permalink raw reply [flat|nested] 8+ messages in thread
* [patch 2.4 1/2] backport 2.6 x86 cpu capabilities
@ 2003-08-08 2:54 Jeff Garzik
0 siblings, 0 replies; 8+ messages in thread
From: Jeff Garzik @ 2003-08-08 2:54 UTC (permalink / raw)
To: linux-kernel
(hopefully destined for 2.4.23-pre1)
#
# include/asm-i386/msr.h 1.8 -> 1.9
# include/asm-i386/cpufeature.h 1.5 -> 1.6
# arch/i386/kernel/setup.c 1.70 -> 1.71
#
# --------------------------------------------
# 03/08/07 jgarzik@redhat.com 1.1066
# [ia32] Via, Intel cpu capabilities update
#
# * /proc/cpuinfo support for Intel Prescott New Instructions (PNI)
# * /proc/cpuinfo support for Centuar Extended Feature Flags
# (including "xstore", Via RNG support)
# * at boot time, input x86 capability data for the two featuresets
# * sync include/asm-i386/cpufeature.h definitions with 2.6.x
# --------------------------------------------
#
diff -Nru a/arch/i386/kernel/setup.c b/arch/i386/kernel/setup.c
--- a/arch/i386/kernel/setup.c Thu Aug 7 22:51:33 2003
+++ b/arch/i386/kernel/setup.c Thu Aug 7 22:51:33 2003
@@ -1965,6 +1965,37 @@
#endif
+static void __init init_c3(struct cpuinfo_x86 *c)
+{
+ u32 lo, hi;
+
+ /* Test for Centaur Extended Feature Flags presence */
+ if (cpuid_eax(0xC0000000) >= 0xC0000001) {
+ /* store Centaur Extended Feature Flags as
+ * word 5 of the CPU capability bit array
+ */
+ c->x86_capability[5] = cpuid_edx(0xC0000001);
+ }
+
+ switch (c->x86_model) {
+ case 6 ... 8: /* Cyrix III family */
+ rdmsr (MSR_VIA_FCR, lo, hi);
+ lo |= (1<<1 | 1<<7); /* Report CX8 & enable PGE */
+ wrmsr (MSR_VIA_FCR, lo, hi);
+
+ set_bit(X86_FEATURE_CX8, c->x86_capability);
+ set_bit(X86_FEATURE_3DNOW, c->x86_capability);
+
+ /* fall through */
+
+ case 9: /* Nehemiah */
+ default:
+ get_model_name(c);
+ display_cacheinfo(c);
+ break;
+ }
+}
+
static void __init init_centaur(struct cpuinfo_x86 *c)
{
enum {
@@ -2103,23 +2134,7 @@
break;
case 6:
- switch (c->x86_model) {
- case 6 ... 8: /* Cyrix III family */
- rdmsr (MSR_VIA_FCR, lo, hi);
- lo |= (1<<1 | 1<<7); /* Report CX8 & enable PGE */
- wrmsr (MSR_VIA_FCR, lo, hi);
-
- set_bit(X86_FEATURE_CX8, &c->x86_capability);
- set_bit(X86_FEATURE_3DNOW, &c->x86_capability);
-
- /* fall through */
-
- case 9: /* Nehemiah */
- default:
- get_model_name(c);
- display_cacheinfo(c);
- break;
- }
+ init_c3(c);
break;
}
}
@@ -2754,10 +2769,16 @@
/* Intel-defined flags: level 0x00000001 */
if ( c->cpuid_level >= 0x00000001 ) {
- cpuid(0x00000001, &tfms, &junk, &junk,
- &c->x86_capability[0]);
+ u32 capability, excap;
+ cpuid(0x00000001, &tfms, &junk, &excap, &capability);
+ c->x86_capability[0] = capability;
+ c->x86_capability[4] = excap;
c->x86 = (tfms >> 8) & 15;
c->x86_model = (tfms >> 4) & 15;
+ if (c->x86 == 0xf) {
+ c->x86 += (tfms >> 20) & 0xff;
+ c->x86_model += ((tfms >> 16) & 0xF) << 4;
+ }
c->x86_mask = tfms & 15;
} else {
/* Have CPUID level 0 only - unheard of */
@@ -2961,12 +2982,12 @@
"fpu", "vme", "de", "pse", "tsc", "msr", "pae", "mce",
"cx8", "apic", NULL, "sep", "mtrr", "pge", "mca", "cmov",
"pat", "pse36", "pn", "clflush", NULL, "dts", "acpi", "mmx",
- "fxsr", "sse", "sse2", "ss", "ht", "tm", "ia64", NULL,
+ "fxsr", "sse", "sse2", "ss", "ht", "tm", "ia64", "pbe",
/* AMD-defined */
NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
NULL, NULL, NULL, "syscall", NULL, NULL, NULL, NULL,
- NULL, NULL, NULL, NULL, NULL, NULL, "mmxext", NULL,
+ NULL, NULL, NULL, "mp", NULL, NULL, "mmxext", NULL,
NULL, NULL, NULL, NULL, NULL, "lm", "3dnowext", "3dnow",
/* Transmeta-defined */
@@ -2976,7 +2997,20 @@
NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
/* Other (Linux-defined) */
- "cxmmx", "k6_mtrr", "cyrix_arr", "centaur_mcr", NULL, NULL, NULL, NULL,
+ "cxmmx", "k6_mtrr", "cyrix_arr", "centaur_mcr",
+ NULL, NULL, NULL, NULL,
+ NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
+ NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
+ NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
+
+ /* Intel-defined (#2) */
+ "pni", NULL, NULL, "monitor", "ds_cpl", NULL, NULL, "tm2",
+ "est", NULL, "cid", NULL, NULL, NULL, NULL, NULL,
+ NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
+ NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
+
+ /* VIA/Cyrix/Centaur-defined */
+ NULL, NULL, "xstore", NULL, NULL, NULL, NULL, NULL,
NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
diff -Nru a/include/asm-i386/cpufeature.h b/include/asm-i386/cpufeature.h
--- a/include/asm-i386/cpufeature.h Thu Aug 7 22:51:33 2003
+++ b/include/asm-i386/cpufeature.h Thu Aug 7 22:51:33 2003
@@ -10,9 +10,9 @@
/* Sample usage: CPU_FEATURE_P(cpu.x86_capability, FPU) */
#define CPU_FEATURE_P(CAP, FEATURE) test_bit(CAP, X86_FEATURE_##FEATURE ##_BIT)
-#define NCAPINTS 4 /* Currently we have 4 32-bit words worth of info */
+#define NCAPINTS 6 /* Currently we have 6 32-bit words worth of info */
-/* Intel-defined CPU features, CPUID level 0x00000001, word 0 */
+/* Intel-defined CPU features, CPUID level 0x00000001 (edx), word 0 */
#define X86_FEATURE_FPU (0*32+ 0) /* Onboard FPU */
#define X86_FEATURE_VME (0*32+ 1) /* Virtual Mode Extensions */
#define X86_FEATURE_DE (0*32+ 2) /* Debugging Extensions */
@@ -47,6 +47,7 @@
/* AMD-defined CPU features, CPUID level 0x80000001, word 1 */
/* Don't duplicate feature flags which are redundant with Intel! */
#define X86_FEATURE_SYSCALL (1*32+11) /* SYSCALL/SYSRET */
+#define X86_FEATURE_MP (1*32+19) /* MP Capable. */
#define X86_FEATURE_MMXEXT (1*32+22) /* AMD MMX extensions */
#define X86_FEATURE_LM (1*32+29) /* Long Mode (x86-64) */
#define X86_FEATURE_3DNOWEXT (1*32+30) /* AMD 3DNow! extensions */
@@ -63,10 +64,19 @@
#define X86_FEATURE_K6_MTRR (3*32+ 1) /* AMD K6 nonstandard MTRRs */
#define X86_FEATURE_CYRIX_ARR (3*32+ 2) /* Cyrix ARRs (= MTRRs) */
#define X86_FEATURE_CENTAUR_MCR (3*32+ 3) /* Centaur MCRs (= MTRRs) */
+/* cpu types for specific tunings: */
+#define X86_FEATURE_K8 (3*32+ 4) /* Opteron, Athlon64 */
+#define X86_FEATURE_K7 (3*32+ 5) /* Athlon */
+#define X86_FEATURE_P3 (3*32+ 6) /* P3 */
+#define X86_FEATURE_P4 (3*32+ 7) /* P4 */
-/* Intel defined CPU features, CPUID level 0x00000001 (ecx), word 4 */
+/* Intel-defined CPU features, CPUID level 0x00000001 (ecx), word 4 */
#define X86_FEATURE_EST (4*32+ 7) /* Enhanced SpeedStep */
+/* VIA/Cyrix/Centaur-defined CPU features, CPUID level 0xC0000001, word 5 */
+#define X86_FEATURE_XSTORE (5*32+ 2) /* on-CPU RNG present (xstore insn) */
+
+
#define cpu_has(c, bit) test_bit(bit, (c)->x86_capability)
#define boot_cpu_has(bit) test_bit(bit, boot_cpu_data.x86_capability)
@@ -77,7 +87,9 @@
#define cpu_has_tsc boot_cpu_has(X86_FEATURE_TSC)
#define cpu_has_pae boot_cpu_has(X86_FEATURE_PAE)
#define cpu_has_pge boot_cpu_has(X86_FEATURE_PGE)
+#define cpu_has_sse2 boot_cpu_has(X86_FEATURE_XMM2)
#define cpu_has_apic boot_cpu_has(X86_FEATURE_APIC)
+#define cpu_has_sep boot_cpu_has(X86_FEATURE_SEP)
#define cpu_has_mtrr boot_cpu_has(X86_FEATURE_MTRR)
#define cpu_has_mmx boot_cpu_has(X86_FEATURE_MMX)
#define cpu_has_fxsr boot_cpu_has(X86_FEATURE_FXSR)
@@ -87,6 +99,7 @@
#define cpu_has_k6_mtrr boot_cpu_has(X86_FEATURE_K6_MTRR)
#define cpu_has_cyrix_arr boot_cpu_has(X86_FEATURE_CYRIX_ARR)
#define cpu_has_centaur_mcr boot_cpu_has(X86_FEATURE_CENTAUR_MCR)
+#define cpu_has_xstore boot_cpu_has(X86_FEATURE_XSTORE)
#endif /* __ASM_I386_CPUFEATURE_H */
diff -Nru a/include/asm-i386/msr.h b/include/asm-i386/msr.h
--- a/include/asm-i386/msr.h Thu Aug 7 22:51:33 2003
+++ b/include/asm-i386/msr.h Thu Aug 7 22:51:33 2003
@@ -113,6 +113,7 @@
/* VIA Cyrix defined MSRs*/
#define MSR_VIA_FCR 0x1107
#define MSR_VIA_LONGHAUL 0x110a
+#define MSR_VIA_RNG 0x110b
#define MSR_VIA_BCR2 0x1147
/* Transmeta defined MSRs */
^ permalink raw reply [flat|nested] 8+ messages in thread
end of thread, other threads:[~2003-08-11 0:29 UTC | newest]
Thread overview: 8+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2003-08-08 19:41 [patch 2.4 1/2] backport 2.6 x86 cpu capabilities Mikael Pettersson
2003-08-09 14:05 ` Jeff Garzik
-- strict thread matches above, loose matches on Subject: below --
2003-08-10 23:59 Albert Cahalan
2003-08-11 0:28 ` Jeff Garzik
2003-08-08 11:19 Mikael Pettersson
2003-08-08 13:15 ` Jeff Garzik
2003-08-09 9:51 ` Marc-Christian Petersen
2003-08-08 2:54 Jeff Garzik
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).