FW: [ia64 R&D] CONFIG_IA64_MCA

FW: [ia64 R&D] CONFIG_IA64_MCA

[John Curry]

Hello Mr. Jesse Barnes,
	could you please revert the CONFIG_MCA removal for the reasons
described below.

					Thanks,
						J. Curry



Oops, I missed that this came in via ia64_rd, not linux-ia64.  It would be
best to send mail to Jesse Barnes to ask him to revert the CONFIG_MCA
removal (with cc to linux-ia64).

Thanks,

	--david

>>>>> On Fri, 30 Apr 2004 16:03:44 -0700, "John Curry" 
>>>>> <jcurry@cup.hp.com> said:

  John> Hi, I have heard rumors that the CONFIG_IA64_MCA option may be
  John> phased out. If this is true, this email is an attempt to lobby
  John> AGAINST such a move.  I work with a team that explicitly
  John> disables the CONFIG_IA64_MCA so that the kernel will NOT
  John> handle MCAs. If kernel handling of MCAs is made automatic,
  John> then that would interfere with our use of the
  John> kernel. Hopefully, this rumor is false.

  John> 						Thanks,
  John> J. Curry

Re: FW: [ia64 R&D] CONFIG_IA64_MCA

[Jesse Barnes]

[-- Attachment #1: Type: text/plain, Size: 648 bytes --]

John> Hi, I have heard rumors that the CONFIG_IA64_MCA option may be
John> phased out. If this is true, this email is an attempt to lobby
John> AGAINST such a move.  I work with a team that explicitly
John> disables the CONFIG_IA64_MCA so that the kernel will NOT
John> handle MCAs. If kernel handling of MCAs is made automatic,
John> then that would interfere with our use of the
John> kernel. Hopefully, this rumor is false.

Too bad you didn't send this message to the linux-ia64 list at the time I 
posted the patch.  Oh well.  Here's the original patch that you can revert 
with 'patch -R'.  I'll put something better together shortly.

Jesse

[-- Attachment #2: kill-config-mca.patch --]
[-- Type: text/plain, Size: 4144 bytes --]

===== arch/ia64/Kconfig 1.63 vs edited =====
--- 1.63/arch/ia64/Kconfig	Thu Mar 11 12:58:54 2004
+++ edited/arch/ia64/Kconfig	Mon Mar 15 14:21:32 2004
@@ -238,13 +238,6 @@
 	  or have huge holes in the physical address space for other reasons.
 	  See <file:Documentation/vm/numa> for more.
 
-config IA64_MCA
-	bool "Machine Check Abort"
-	default y if !IA64_HP_SIM
-	help
-	  Say Y here to enable machine check support for IA-64.  If you're
-	  unsure, answer Y.
-
 config IA64_CYCLONE
 	bool "Support Cyclone(EXA) Time Source"
 	help
===== arch/ia64/kernel/Makefile 1.29 vs edited =====
--- 1.29/arch/ia64/kernel/Makefile	Mon Feb 23 17:44:56 2004
+++ edited/arch/ia64/kernel/Makefile	Mon Mar 15 14:17:01 2004
@@ -6,13 +6,12 @@
 
 obj-y := acpi.o entry.o efi.o efi_stub.o gate-data.o fsys.o ia64_ksyms.o irq.o irq_ia64.o	\
 	 irq_lsapic.o ivt.o machvec.o pal.o patch.o process.o perfmon.o ptrace.o sal.o		\
-	 salinfo.o semaphore.o setup.o signal.o sys_ia64.o time.o traps.o unaligned.o unwind.o
+	 salinfo.o semaphore.o setup.o signal.o sys_ia64.o time.o traps.o unaligned.o unwind.o mca.o mca_asm.o
 
 obj-$(CONFIG_EFI_VARS)		+= efivars.o
 obj-$(CONFIG_IA64_BRL_EMU)	+= brl_emu.o
 obj-$(CONFIG_IA64_GENERIC)	+= acpi-ext.o
 obj-$(CONFIG_IA64_HP_ZX1)	+= acpi-ext.o
-obj-$(CONFIG_IA64_MCA)		+= mca.o mca_asm.o
 obj-$(CONFIG_IA64_PALINFO)	+= palinfo.o
 obj-$(CONFIG_IOSAPIC)		+= iosapic.o
 obj-$(CONFIG_MODULES)		+= module.o
===== arch/ia64/kernel/efi.c 1.29 vs edited =====
--- 1.29/arch/ia64/kernel/efi.c	Mon Jan  5 07:43:05 2004
+++ edited/arch/ia64/kernel/efi.c	Mon Mar 15 14:20:10 2004
@@ -399,9 +399,7 @@
 	int pal_code_count = 0;
 	u64 mask, psr;
 	u64 vaddr;
-#ifdef CONFIG_IA64_MCA
 	int cpu;
-#endif
 
 	efi_map_start = __va(ia64_boot_param->efi_memmap);
 	efi_map_end   = efi_map_start + ia64_boot_param->efi_memmap_size;
@@ -463,13 +461,11 @@
 		ia64_set_psr(psr);		/* restore psr */
 		ia64_srlz_i();
 
-#ifdef CONFIG_IA64_MCA
 		cpu = smp_processor_id();
 
 		/* insert this TR into our list for MCA recovery purposes */
 		ia64_mca_tlb_list[cpu].pal_base = vaddr & mask;
 		ia64_mca_tlb_list[cpu].pal_paddr = pte_val(mk_pte_phys(md->phys_addr, PAGE_KERNEL));
-#endif
 	}
 }
 
===== arch/ia64/kernel/setup.c 1.68 vs edited =====
--- 1.68/arch/ia64/kernel/setup.c	Mon Jan 19 15:38:10 2004
+++ edited/arch/ia64/kernel/setup.c	Mon Mar 15 14:20:19 2004
@@ -355,10 +355,8 @@
 # endif
 #endif
 
-#ifdef CONFIG_IA64_MCA
 	/* enable IA-64 Machine Check Abort Handling */
 	ia64_mca_init();
-#endif
 
 	platform_setup(cmdline_p);
 	paging_init();
===== arch/ia64/kernel/smpboot.c 1.47 vs edited =====
--- 1.47/arch/ia64/kernel/smpboot.c	Thu Feb 12 16:54:17 2004
+++ edited/arch/ia64/kernel/smpboot.c	Mon Mar 15 14:20:27 2004
@@ -300,9 +300,7 @@
 	 */
 	ia64_set_kr(IA64_KR_IO_BASE, __pa(ia64_iobase));
 
-#ifdef CONFIG_IA64_MCA
 	ia64_mca_cmc_vector_setup();	/* Setup vector on AP & enable */
-#endif
 
 #ifdef CONFIG_PERFMON
 	pfm_init_percpu();
===== arch/ia64/mm/init.c 1.56 vs edited =====
--- 1.56/arch/ia64/mm/init.c	Thu Feb 26 12:18:51 2004
+++ edited/arch/ia64/mm/init.c	Mon Mar 15 14:20:43 2004
@@ -279,9 +279,7 @@
 {
 	unsigned long psr, pta, impl_va_bits;
 	extern void __init tlb_init (void);
-#ifdef CONFIG_IA64_MCA
 	int cpu;
-#endif
 
 #ifdef CONFIG_DISABLE_VHPT
 #	define VHPT_ENABLE_BIT	0
@@ -346,7 +344,6 @@
 	ia64_set_rr(HPAGE_REGION_BASE, HPAGE_SHIFT << 2);
 #endif
 
-#ifdef	CONFIG_IA64_MCA
 	cpu = smp_processor_id();
 
 	/* mca handler uses cr.lid as key to pick the right entry */
@@ -360,7 +357,6 @@
 	ia64_mca_tlb_list[cpu].ptce_count[1] = local_cpu_data->ptce_count[1];
 	ia64_mca_tlb_list[cpu].ptce_stride[0] = local_cpu_data->ptce_stride[0];
 	ia64_mca_tlb_list[cpu].ptce_stride[1] = local_cpu_data->ptce_stride[1];
-#endif
 }
 
 #ifdef CONFIG_VIRTUAL_MEM_MAP
===== arch/ia64/sn/kernel/setup.c 1.33 vs edited =====
--- 1.33/arch/ia64/sn/kernel/setup.c	Sat Feb 14 23:22:40 2004
+++ edited/arch/ia64/sn/kernel/setup.c	Mon Mar 15 14:19:50 2004
@@ -194,10 +194,7 @@
 	}
 }
 
-#ifdef CONFIG_IA64_MCA
 extern int platform_intr_list[];
-#endif
-
 extern nasid_t master_nasid;
 static int shub_1_1_found __initdata;
 

RE: [ia64 R&D] CONFIG_IA64_MCA

[Luck, Tony]

John> Hi, I have heard rumors that the CONFIG_IA64_MCA option may be
John> phased out. If this is true, this email is an attempt to lobby
John> AGAINST such a move.  I work with a team that explicitly
John> disables the CONFIG_IA64_MCA so that the kernel will NOT
John> handle MCAs. If kernel handling of MCAs is made automatic,
John> then that would interfere with our use of the
John> kernel. Hopefully, this rumor is false.

Would a boot option to skip registration linux handlers with
SAL work for you?  That would allow us to keep Jesse's cleanup
that removes a bunch of #ifdefs.

-Tony

Re: [ia64 R&D] CONFIG_IA64_MCA

[Jesse Barnes]

On Tuesday, May 4, 2004 6:53 am, Luck, Tony wrote:
> John> Hi, I have heard rumors that the CONFIG_IA64_MCA option may be
> John> phased out. If this is true, this email is an attempt to lobby
> John> AGAINST such a move.  I work with a team that explicitly
> John> disables the CONFIG_IA64_MCA so that the kernel will NOT
> John> handle MCAs. If kernel handling of MCAs is made automatic,
> John> then that would interfere with our use of the
> John> kernel. Hopefully, this rumor is false.
>
> Would a boot option to skip registration linux handlers with
> SAL work for you?  That would allow us to keep Jesse's cleanup
> that removes a bunch of #ifdefs.

That seems like an even better option, since it means they wouldn't have to 
recompile their kernel (unless they're making other changes), and may be 
useful in general.

Jesse

RE: [ia64 R&D] CONFIG_IA64_MCA

[CURRY,JOHN W (HP-Cupertino,ex1)]

Yes, I think that could work. Tell me more. I have already communicated with
Jesse, though.
			Thanks,
				J. Curry

-----Original Message-----
From: linux-ia64-owner@vger.kernel.org
[mailto:linux-ia64-owner@vger.kernel.org] On Behalf Of Luck, Tony
Sent: Tuesday, May 04, 2004 6:53 AM
To: John Curry; jbarnes@sgi.com
Cc: linux-ia64@vger.kernel.org
Subject: RE: [ia64 R&D] CONFIG_IA64_MCA


John> Hi, I have heard rumors that the CONFIG_IA64_MCA option may be 
John> phased out. If this is true, this email is an attempt to lobby 
John> AGAINST such a move.  I work with a team that explicitly disables 
John> the CONFIG_IA64_MCA so that the kernel will NOT handle MCAs. If 
John> kernel handling of MCAs is made automatic, then that would 
John> interfere with our use of the kernel. Hopefully, this rumor is 
John> false.

Would a boot option to skip registration linux handlers with SAL work for
you?  That would allow us to keep Jesse's cleanup that removes a bunch of
#ifdefs.

-Tony
-
To unsubscribe from this list: send the line "unsubscribe linux-ia64" in the
body of a message to majordomo@vger.kernel.org More majordomo info at
http://vger.kernel.org/majordomo-info.html

Re: [ia64 R&D] CONFIG_IA64_MCA

[Jesse Barnes]

[-- Attachment #1: Type: text/plain, Size: 546 bytes --]

On Tuesday, May 4, 2004 9:06 am, CURRY,JOHN W (HP-Cupertino,ex1) wrote:
> Yes, I think that could work. Tell me more. I have already communicated
> with Jesse, though.

It might look something like this (though now that I've looked at it, it seems 
we could use an early option parser for ia64).  I've tried to add a 'nomca' 
boot option that you can pass to the kernel to avoid registering the MCA 
handlers with the firmware.  This kernel builds and boots for me w/o the 
'nomca' option, but on sn2 it hangs at boot if I pass 'nomca'.

Jesse



[-- Attachment #2: no-mca-register.patch --]
[-- Type: text/plain, Size: 832 bytes --]

===== arch/ia64/kernel/setup.c 1.70 vs edited =====
--- 1.70/arch/ia64/kernel/setup.c	Wed Mar 17 04:46:59 2004
+++ edited/arch/ia64/kernel/setup.c	Tue May  4 09:23:34 2004
@@ -283,6 +283,9 @@
 void __init
 setup_arch (char **cmdline_p)
 {
+	char *cp;
+	int nomca = 0;
+
 	unw_init();
 
 	ia64_patch_vtop((u64) __start___vtop_patchlist, (u64) __end___vtop_patchlist);
@@ -359,9 +362,19 @@
 		conswitchp = &vga_con;
 # endif
 #endif
-
-	/* enable IA-64 Machine Check Abort Handling */
-	ia64_mca_init();
+	for (cp = saved_command_line; *cp; )
+		if (memcmp(cp, "nomca", 5) == 0)
+			nomca = 1;
+		else {
+			while (*cp != ' ' && *cp)
+				++cp;
+			while (*cp == ' ')
+				++cp;
+		}
+
+	/* enable IA-64 Machine Check Abort Handling unless disabled */
+	if (!nomca)
+		ia64_mca_init();
 
 	platform_setup(cmdline_p);
 	paging_init();

RE: [ia64 R&D] CONFIG_IA64_MCA

[CURRY,JOHN W (HP-Cupertino,ex1)]

Well, once it's working I think this will be okay for our team.
		Thanks,
			J. Curry

-----Original Message-----
From: Jesse Barnes [mailto:jbarnes@engr.sgi.com] 
Sent: Tuesday, May 04, 2004 9:34 AM
To: CURRY,JOHN W (HP-Cupertino,ex1)
Cc: 'Luck, Tony'; linux-ia64@vger.kernel.org
Subject: Re: [ia64 R&D] CONFIG_IA64_MCA


On Tuesday, May 4, 2004 9:06 am, CURRY,JOHN W (HP-Cupertino,ex1) wrote:
> Yes, I think that could work. Tell me more. I have already 
> communicated with Jesse, though.
It might look something like this (though now that I've looked at it, it
seems 
we could use an early option parser for ia64).  I've tried to add a 'nomca' 
boot option that you can pass to the kernel to avoid registering the MCA 
handlers with the firmware.  This kernel builds and boots for me w/o the 
'nomca' option, but on sn2 it hangs at boot if I pass 'nomca'. Jesse

Re: [ia64 R&D] CONFIG_IA64_MCA

[Jesse Barnes]

On Tuesday, May 4, 2004 7:51 pm, CURRY,JOHN W (HP-Cupertino,ex1) wrote:
> Well, once it's working I think this will be okay for our team.

Any chance you could test the patch for me?  Anyone else have hw they could 
try it on?

Thanks,
Jesse

RE: [ia64 R&D] CONFIG_IA64_MCA

[CURRY,JOHN W (HP-Cupertino,ex1)]

Sure,
	But I won't be able to get to it till sometime between May 10 and
May 14.

						Thanks,
							J. Curry

-----Original Message-----
From: Jesse Barnes [mailto:jbarnes@engr.sgi.com] 
Sent: Friday, May 07, 2004 12:57 PM
To: CURRY,JOHN W (HP-Cupertino,ex1)
Cc: 'Luck, Tony'; 'linux-ia64@vger.kernel.org'
Subject: Re: [ia64 R&D] CONFIG_IA64_MCA


On Tuesday, May 4, 2004 7:51 pm, CURRY,JOHN W (HP-Cupertino,ex1) wrote:
> Well, once it's working I think this will be okay for our team.

Any chance you could test the patch for me?  Anyone else have hw they could 
try it on?

Thanks,
Jesse

RE: [ia64 R&D] CONFIG_IA64_MCA

[Chen, Kenneth W]

>>>> Jesse Barnes wrote on Friday, May 07, 2004 12:57 PM
>
> On Tuesday, May 4, 2004 7:51 pm, CURRY,JOHN W (HP-Cupertino,ex1) wrote:
> > Well, once it's working I think this will be okay for our team.
>
> Any chance you could test the patch for me?  Anyone else have hw they
> could try it on?

Is there another patch or this is the latest one?
http://www.gelato.unsw.edu.au/linux-ia64/0405/9514.html

If that's the latest one, then it won't boot on any machine with nomca
parameter because once match found, it stuck in the for loop and memcmp
endlessly, cp never gets incremented :-(

How about this one:

diff -Nur linux-2.6.6-rc3/arch/ia64/kernel/setup.c linux-2.6.6-rc3.ken/arch/ia64/kernel/setup.c
--- linux-2.6.6-rc3/arch/ia64/kernel/setup.c	2004-04-27 18:35:09.000000000 -0700
+++ linux-2.6.6-rc3.ken/arch/ia64/kernel/setup.c	2004-05-07 14:22:18.000000000 -0700
@@ -360,8 +360,9 @@
 # endif
 #endif

-	/* enable IA-64 Machine Check Abort Handling */
-	ia64_mca_init();
+	/* enable IA-64 Machine Check Abort Handling unless disabled */
+	if (!strstr(saved_command_line, "nomca "))
+		ia64_mca_init();

 	platform_setup(cmdline_p);
 	paging_init();

Re: [ia64 R&D] CONFIG_IA64_MCA

[Jesse Barnes]

On Friday, May 7, 2004 2:28 pm, Chen, Kenneth W wrote:
> Is there another patch or this is the latest one?
> http://www.gelato.unsw.edu.au/linux-ia64/0405/9514.html
>
> If that's the latest one, then it won't boot on any machine with nomca
> parameter because once match found, it stuck in the for loop and memcmp
> endlessly, cp never gets incremented :-(

Oh, oops.  Your patch looks simpler (and correct!) :)

Jesse

RE: [ia64 R&D] CONFIG_IA64_MCA

[Sourav Sen]

Sorry to dig thigs up from the past, but when I apply
this against 2.6.6 and pass the nomca flag, I get MCA in 
the boot path on sx1000 based systems of HP. 

Thanks
Sourav
HP-STS, Bangalore
+91-80-2205-1897


+ -----Original Message-----
+ From: linux-ia64-owner@vger.kernel.org
+ [mailto:linux-ia64-owner@vger.kernel.org]On Behalf Of Chen, Kenneth W
+ Sent: Saturday, May 08, 2004 2:58 AM
+ To: 'Jesse Barnes'; CURRY,JOHN W (HP-Cupertino,ex1)
+ Cc: Luck, Tony; linux-ia64@vger.kernel.org
+ Subject: RE: [ia64 R&D] CONFIG_IA64_MCA
+ 
+ 
+ >>>> Jesse Barnes wrote on Friday, May 07, 2004 12:57 PM
+ >
+ > On Tuesday, May 4, 2004 7:51 pm, CURRY,JOHN W 
+ (HP-Cupertino,ex1) wrote:
+ > > Well, once it's working I think this will be okay for our team.
+ >
+ > Any chance you could test the patch for me?  Anyone else 
+ have hw they
+ > could try it on?
+ 
+ Is there another patch or this is the latest one?
+ http://www.gelato.unsw.edu.au/linux-ia64/0405/9514.html
+ 
+ If that's the latest one, then it won't boot on any machine with nomca
+ parameter because once match found, it stuck in the for loop 
+ and memcmp
+ endlessly, cp never gets incremented :-(
+ 
+ How about this one:
+ 
+ diff -Nur linux-2.6.6-rc3/arch/ia64/kernel/setup.c 
+ linux-2.6.6-rc3.ken/arch/ia64/kernel/setup.c
+ --- linux-2.6.6-rc3/arch/ia64/kernel/setup.c	2004-04-27 
+ 18:35:09.000000000 -0700
+ +++ linux-2.6.6-rc3.ken/arch/ia64/kernel/setup.c	
+ 2004-05-07 14:22:18.000000000 -0700
+ @@ -360,8 +360,9 @@
+  # endif
+  #endif
+ 
+ -	/* enable IA-64 Machine Check Abort Handling */
+ -	ia64_mca_init();
+ +	/* enable IA-64 Machine Check Abort Handling unless disabled */
+ +	if (!strstr(saved_command_line, "nomca "))
+ +		ia64_mca_init();
+ 
+  	platform_setup(cmdline_p);
+  	paging_init();
+ 
+ 
+ -
+ To unsubscribe from this list: send the line "unsubscribe 
+ linux-ia64" in
+ the body of a message to majordomo@vger.kernel.org
+ More majordomo info at  http://vger.kernel.org/majordomo-info.html
+ 

Re: [ia64 R&D] CONFIG_IA64_MCA

[Jesse Barnes]

On Tuesday, August 3, 2004 6:45 am, Sourav Sen wrote:
> Sorry to dig thigs up from the past, but when I apply
> this against 2.6.6 and pass the nomca flag, I get MCA in
> the boot path on sx1000 based systems of HP.

Yeah, I think Ken pointed this out shortly after I posted the patch.  I 
thought a fixed version made its way into the tree?  Tony?

Thanks,
Jesse

RE: [ia64 R&D] CONFIG_IA64_MCA

[Luck, Tony]

Jesse Barnes wrote:
>On Tuesday, August 3, 2004 6:45 am, Sourav Sen wrote:
>> Sorry to dig thigs up from the past, but when I apply
>> this against 2.6.6 and pass the nomca flag, I get MCA in
>> the boot path on sx1000 based systems of HP.
>
>Yeah, I think Ken pointed this out shortly after I posted the 
>patch.  I thought a fixed version made its way into the tree?  Tony?

I don't see any traces of it in the tree.

I think that the old version of the patch doesn't work with
the current tree because just skipping ia64_mca_init() won't
work now.  You'd still call mca_late_init() which would do
a few bad things not realizing that ia64_mca_init() hadn't
been called to prepare stuff.

-Tony