Skip to contentSkip to conversation

Finally: success with 2.4.2-ia64-010228

Linux-ia64 Keith Owens 1 participant 1 message Mar 16, 2001 ACPI

lore
Share & tools
lore
#0 lore raw

Several independent bugs combined to make 2.4.2-ia64-010228 fail. The first two patches below came from an SGI mailing list and may already be known on this list, although I have not seen them mentioned here.

The ivt.S change caused random errors on kdb break points.

The missing '\' in ia64/config.in slips through menuconfig and results in CONFIG_ITANIUM_PTCG being set to y for B1 and B2. As soon as init calls mprotect() on my B1 it does ptcg and hangs. This explains why some people were reporting hang after the kernel gives control to user space.

I am still seeing random EFI delays, before it even gets to the kernel, using build 84. AFAICT there is no higher build for B1.

Index: 2.17/drivers/acpi/resources/rscalc.c

drivers/acpi/resources/rscalc.c Fri, 16 Mar 2001 13:31:09 +1100 kaos (linux-2.4/q/b/43_rscalc.c 1.3 644) renamed from drivers/acpi/resources/rscalc.c Sat, 10 Feb 2001 13:20:52 +1100 kaos (linux-2.4/q/b/43_rscalc.c 1.3 644) +1 -1
@@ -866,7 +866,7 @@ acpi_rs_calculate_pci_routing_table_leng
 	}
 
 
-	*buffer_size_needed = temp_size_needed;
+	*buffer_size_needed = temp_size_needed + sizeof(PCI_ROUTING_TABLE);
 
 	return (AE_OK);
 }

Index: 2.17/drivers/acpi/common/cmxface.c

drivers/acpi/common/cmxface.c Fri, 16 Mar 2001 13:33:07 +1100 kaos (linux-2.4/s/b/41_cmxface.c 1.4 644) renamed from drivers/acpi/common/cmxface.c Sat, 10 Feb 2001 13:20:52 +1100 kaos (linux-2.4/s/b/41_cmxface.c 1.4 644) +5 -0
@@ -158,7 +158,12 @@ acpi_enable_subsystem (
 	if (!(flags & ACPI_NO_ACPI_ENABLE)) {
 		status = acpi_enable ();
 		if (ACPI_FAILURE (status)) {
+#define BIGSUR_WA
+#ifdef BIGSUR_WA
+			printk("acpi_enable fail:ignored\n");
+#else
 			return (status);
+#endif
 		}
 
 	/*

* Note:
Index: 2.17/arch/ia64/kernel/ivt.S

arch/ia64/kernel/ivt.S Fri, 16 Mar 2001 15:05:27 +1100 kaos (linux-2.4/s/c/6_ivt.S 1.1.2.1 644) renamed from arch/ia64/kernel/ivt.S Fri, 02 Mar 2001 13:03:52 +1100 kaos (linux-2.4/s/c/6_ivt.S 1.1.2.1 644) +1 -1
@@ -899,7 +899,7 @@ ENTRY(non_syscall)
 	// suitable spot...
 
 	alloc r14=ar.pfs,0,0,2,0
-	mov out0=r8
+	mov out0=cr.iim			// get break immediate (must be done while psr.ic is off)
 	add out1,sp
 	adds r3=8,r2			// set up second base pointer for SAVE_REST
 

Index: 2.17/arch/ia64/config.in

arch/ia64/config.in Fri, 16 Mar 2001 17:47:20 +1100 kaos (linux-2.4/s/c/38_config.in 1.1.2.1.1.2 644) renamed from arch/ia64/config.in Fri, 02 Mar 2001 23:28:22 +1100 kaos (linux-2.4/s/c/38_config.in 1.1.2.1.1.2 644) +1 -1
@@ -57,7 +57,7 @@ if [ "$CONFIG_ITANIUM" = "y" ]; then
 	if [ "$CONFIG_ITANIUM_CSTEP_SPECIFIC" = "y" ]; then
 	  bool '   Enable Itanium C0-step specific code' CONFIG_ITANIUM_C0_SPECIFIC
 	fi
-	if [ "$CONFIG_ITANIUM_ASTEP_SPECIFIC" = "y" -o "$CONFIG_ITANIUM_B0_SPECIFIC" = "y"
+	if [ "$CONFIG_ITANIUM_ASTEP_SPECIFIC" = "y" -o "$CONFIG_ITANIUM_B0_SPECIFIC" = "y" \
 	     -o "$CONFIG_ITANIUM_B1_SPECIFIC" = "y" -o "$CONFIG_ITANIUM_B2_SPECIFIC" = "y" ]; then
 	  define_bool CONFIG_ITANIUM_PTCG n
 	else