Skip to contentSkip to conversation

cell bug fixes for 2.6.20

PATCH In review Arnd Bergmann 3 participants 15 messages Nov 27, 2006 – Nov 28, 2006

Subsystems (4)
lore
Share & tools
lore
#0 lore raw

cover cell bug fixes for 2.6.20

This seems to be the fallout of merging so many large patches in the recent times, mostly stuff broke by combining recent features.

If everyone likes these patches, please pull them from kernel.org/pub/scm/linux/kernel/git/arnd/cell-2.6.git cell-merge

	Arnd <><

--

#1 lore raw

PATCH 1/9 spufs: return an error in spu_create is isolated create isnt supported

From: Jeremy Kerr <[email protected]>
This changes the spu_create system call to return an error (-ENODEV) if and isolated spu context is requested on hardware that doesn't support isolated mode.

Tested on systemsim with and without isolation support

arch/powerpc/platforms/cell/spufs/inode.c +4 -0
Index: linux-2.6/arch/powerpc/platforms/cell/spufs/inode.c
===================================================================
@@ -323,6 +323,10 @@ static int spufs_create_context(struct i
 	    == SPU_CREATE_ISOLATE)
 		goto out_unlock;
 
+	ret = -ENODEV;
+	if ((flags & SPU_CREATE_ISOLATE) && !isolated_loader)
+		goto out_unlock;
+
 	ret = spufs_mkdir(inode, dentry, flags, mode & S_IRWXUGO);
 	if (ret)
 		goto out_unlock;

--

#2 lore raw

PATCH 2/9 spufs: always send sigtrap on breakpoint

In reply to Arnd Bergmann

Currently, we only send a sigtrap if the current task is being ptraced. This is somewhat inconsistant, and it breaks utrace support in fedora. Removing the check should do the right thing in all cases.

arch/powerpc/platforms/cell/spufs/run.c +4 -6
index 1be4e33..1acc2ff 100644
@@ -350,12 +350,10 @@ out2:
 	       (status >> SPU_STOP_STATUS_SHIFT != 0x2104)))))
 		ret = status;
 
-	if (unlikely(current->ptrace & PT_PTRACED)) {
-		if ((status & SPU_STATUS_STOPPED_BY_STOP)
-		    && (status >> SPU_STOP_STATUS_SHIFT) == 0x3fff) {
-			force_sig(SIGTRAP, current);
-			ret = -ERESTARTSYS;
-		}
+	if ((status & SPU_STATUS_STOPPED_BY_STOP)
+	    && (status >> SPU_STOP_STATUS_SHIFT) == 0x3fff) {
+		force_sig(SIGTRAP, current);
+		ret = -ERESTARTSYS;
 	}
 
 out:

--

#3 lore raw

Re: [POWERPC 2/9] spufs: always send sigtrap on breakpoint

Quoted 21 lines · Arnd Bergmann

plain text document attachment (spufs-always-sigtrap-on-breakpoint.diff)
Currently, we only send a sigtrap if the current task is being ptraced. This is somewhat inconsistant, and it breaks utrace support in fedora. Removing the check should do the right thing in all cases.

arch/powerpc/platforms/cell/spufs/run.c +0 -3
index 1be4e33..1acc2ff 100644
@@ -350,12 +350,10 @@ out2:
 	       (status >> SPU_STOP_STATUS_SHIFT != 0x2104)))))
 		ret = status;
 
-	if (unlikely(current->ptrace & PT_PTRACED)) {
-		if ((status & SPU_STATUS_STOPPED_BY_STOP)
-		    && (status >> SPU_STOP_STATUS_SHIFT) == 0x3fff) {

arch/powerpc/platforms/cell/spufs/run.c out2: · line 359

Quoted 10 lines
arch/powerpc/platforms/cell/spufs/run.c +4 -3
-			force_sig(SIGTRAP, current);
-			ret = -ERESTARTSYS;
-		}
+	if ((status & SPU_STATUS_STOPPED_BY_STOP)
+	    && (status >> SPU_STOP_STATUS_SHIFT) == 0x3fff) {
+		force_sig(SIGTRAP, current);
+		ret = -ERESTARTSYS;
 	}
 
 out:

Arnd, could you explain this change a little bit more. Don't you have to consider whether the program is setup to receive SIGTRAP. We know it is code to handle SIGTRAP if it is ptraced. Or, are you asserting that at this point the spu has hit a trap instruction? What does 0x3fff mean?

Luke

#4 lore raw

Re: [POWERPC 2/9] spufs: always send sigtrap on breakpoint

Quoted 6 lines · Luke Browning

=20
Arnd, could you explain this change a little bit more. =A0Don't you have to consider whether the program is setup to receive SIGTRAP. =A0We know it is code to handle SIGTRAP if it is ptraced. =A0Or, are you asserting that at this point the spu has hit a trap instruction? =A0What does 0x3fff mean?

0x3fff is indeed the code returned from the breakpoint instruction, as specified in the JSRE list of stop-and-signal codes. The behaviour is consistant with that of a Linux process entering a breakpoint. This usually only happens while ptracing a task, but there is no reason to treat a task differently when it's not being traced.

It could even be used as well-defined way for an application to create a core dump, similar to abort() in posix.

	Arnd <><
#5 lore raw

PATCH 3/9 spufs: we should only execute init_spu_base on cell

In reply to Arnd Bergmann

From: Stephen Rothwell <[email protected]>

arch/powerpc/platforms/cell/spu_base.c +3 -0
index 841ed35..bd7bffc 100644
@@ -663,6 +663,9 @@ static int __init init_spu_base(void)
 {
 	int i, ret;
 
+	if (!spu_management_ops)
+		return 0;
+
 	/* create sysdev class for spus */
 	ret = sysdev_class_register(&spu_sysdev_class);
 	if (ret)
Signature · 3 lines
1.4.4

--
#6 lore raw

PATCH 4/9 cell: fix building without spufs

In reply to Arnd Bergmann

It may be desireable to build a kernel for cell without spufs, e.g. as the initial kboot kernel. This requires that the SPU specific parts of the core dump and the xmon code depend on CONFIG_SPU_BASE instead of CONFIG_PPC_CELL.

arch/powerpc/xmon/Makefile +1 -1
Index: linux-2.6/arch/powerpc/xmon/Makefile
===================================================================
@@ -8,5 +8,5 @@ obj-y			+= xmon.o setjmp.o start.o nonst
 
 ifdef CONFIG_XMON_DISASSEMBLY
 obj-y			+= ppc-dis.o ppc-opc.o
-obj-$(CONFIG_PPC_CELL)	+= spu-dis.o spu-opc.o
+obj-$(CONFIG_SPU_BASE)	+= spu-dis.o spu-opc.o
 endif
arch/powerpc/xmon/xmon.c +3 -3
Index: linux-2.6/arch/powerpc/xmon/xmon.c
===================================================================
@@ -213,7 +213,7 @@ Commands:\n\
   p 	call a procedure\n\
   r	print registers\n\
   s	single step\n"
-#ifdef CONFIG_PPC_CELL
+#ifdef CONFIG_SPU_BASE
 "  ss	stop execution on all spus\n\
   sr	restore execution on stopped spus\n\
   sf  #	dump spu fields for spu # (in hex)\n\
@@ -2654,7 +2654,7 @@ void __init xmon_setup(void)
 		debugger(NULL);
 }
 
-#ifdef CONFIG_PPC_CELL
+#ifdef CONFIG_SPU_BASE
 
 struct spu_info {
 	struct spu *spu;
@@ -2907,7 +2907,7 @@ static int do_spu_cmd(void)
 
 	return 0;
 }
-#else /* ! CONFIG_PPC_CELL */
+#else /* ! CONFIG_SPU_BASE */
 static int do_spu_cmd(void)
 {
 	return -1;
include/asm-powerpc/elf.h +1 -1
Index: linux-2.6/include/asm-powerpc/elf.h
===================================================================
@@ -411,7 +411,7 @@ do {									\
 /* Keep this the last entry.  */
 #define R_PPC64_NUM		107
 
-#ifdef CONFIG_PPC_CELL
+#ifdef CONFIG_SPU_BASE
 /* Notes used in ET_CORE. Note name is "SPU/<fd>/<filename>". */
 #define NT_SPU		1
 

--

#7 lore raw

PATCH 5/9 ps3: Missed renames of CONFIG_PS3 to CONFIG_PPC_PS3

In reply to Arnd Bergmann

From: Geert Uytterhoeven <[email protected]> When renaming CONFIG_PS3 to CONFIG_PPC_PS3, a few occurrences have been missed.

I also fixed up the alignment in arch/powerpc/platforms/Makefile.

arch/powerpc/platforms/Makefile +3 -3
Index: ps3-linux-src/arch/powerpc/platforms/Makefile
===================================================================
@@ -5,7 +5,7 @@
 obj-$(CONFIG_PPC_PMAC)		+= powermac/
 endif
 endif
-obj-$(CONFIG_PPC_EFIKA)          += efika/
+obj-$(CONFIG_PPC_EFIKA)		+= efika/
 obj-$(CONFIG_PPC_CHRP)		+= chrp/
 obj-$(CONFIG_4xx)		+= 4xx/
 obj-$(CONFIG_PPC_83xx)		+= 83xx/
@@ -14,7 +14,7 @@
 obj-$(CONFIG_PPC_PSERIES)	+= pseries/
 obj-$(CONFIG_PPC_ISERIES)	+= iseries/
 obj-$(CONFIG_PPC_MAPLE)		+= maple/
-obj-$(CONFIG_PPC_PASEMI)		+= pasemi/
+obj-$(CONFIG_PPC_PASEMI)	+= pasemi/
 obj-$(CONFIG_PPC_CELL)		+= cell/
-obj-$(CONFIG_PS3)		+= ps3/
+obj-$(CONFIG_PPC_PS3)		+= ps3/
 obj-$(CONFIG_EMBEDDED6xx)	+= embedded6xx/
drivers/Makefile +1 -1
Index: ps3-linux-src/drivers/Makefile
===================================================================
@@ -77,4 +77,4 @@
 obj-$(CONFIG_SUPERH)		+= sh/
 obj-$(CONFIG_GENERIC_TIME)	+= clocksource/
 obj-$(CONFIG_DMA_ENGINE)	+= dma/
-obj-$(CONFIG_PS3)		+= ps3/
+obj-$(CONFIG_PPC_PS3)		+= ps3/
include/asm-powerpc/firmware.h +2 -2
Index: ps3-linux-src/include/asm-powerpc/firmware.h
===================================================================
@@ -67,7 +67,7 @@
 #ifdef CONFIG_PPC_ISERIES
 		FW_FEATURE_ISERIES_POSSIBLE |
 #endif
-#ifdef CONFIG_PS3
+#ifdef CONFIG_PPC_PS3
 		FW_FEATURE_PS3_POSSIBLE |
 #endif
 		0,
@@ -78,7 +78,7 @@
 #ifdef CONFIG_PPC_ISERIES
 		FW_FEATURE_ISERIES_ALWAYS &
 #endif
-#ifdef CONFIG_PS3
+#ifdef CONFIG_PPC_PS3
 		FW_FEATURE_PS3_ALWAYS &
 #endif
 		FW_FEATURE_POSSIBLE,

Gr{oetje,eeting}s,

Geert

--
Geert Uytterhoeven -- Sony Network and Software Technology Center Europe (NSCE) [email protected] ------- The Corporate Village, Da Vincilaan 7-D1 Voice +32-2-7008453 Fax +32-2-7008622 ---------------- B-1935 Zaventem, Belgium

--

#9 lore raw

PATCH 6/9 ps3: multiplatform build fixes

In reply to Arnd Bergmann

A few code paths need to check whether or not they are running on the PS3's LV1 hypervisor before making hcalls. This introduces a new firmware feature bit for this, FW_FEATURE_PS3_LV1.

Now when both PS3 and IBM_CELL_BLADE are enabled, but not PSERIES, FW_FEATURE_PS3_LV1 and FW_FEATURE_LPAR get enabled at compile time, which is a bug. The same problem can also happen for (PPC_ISERIES && !PPC_PSERIES && PPC_SOMETHING_ELSE). In order to solve this, I introduce a new CONFIG_PPC_NATIVE option that is set when at least one platform is selected that can run without a hypervisor and then turns the firmware feature check into a run-time option.

The new cell oprofile support that was recently merged does not work on hypervisor based platforms like the PS3, therefore make it depend on PPC_CELL_NATIVE instead of PPC_CELL. This may change if we get oprofile support for PS3.

include/asm-powerpc/firmware.h +11 -2
Index: linux-2.6/include/asm-powerpc/firmware.h
===================================================================
@@ -42,6 +42,7 @@
 #define FW_FEATURE_SPLPAR	ASM_CONST(0x0000000000100000)
 #define FW_FEATURE_ISERIES	ASM_CONST(0x0000000000200000)
 #define FW_FEATURE_LPAR		ASM_CONST(0x0000000000400000)
+#define FW_FEATURE_PS3_LV1	ASM_CONST(0x0000000000800000)
 
 #ifndef __ASSEMBLY__
 
@@ -58,8 +59,10 @@ enum {
 	FW_FEATURE_PSERIES_ALWAYS = 0,
 	FW_FEATURE_ISERIES_POSSIBLE = FW_FEATURE_ISERIES | FW_FEATURE_LPAR,
 	FW_FEATURE_ISERIES_ALWAYS = FW_FEATURE_ISERIES | FW_FEATURE_LPAR,
-	FW_FEATURE_PS3_POSSIBLE = FW_FEATURE_LPAR,
-	FW_FEATURE_PS3_ALWAYS = FW_FEATURE_LPAR,
+	FW_FEATURE_PS3_POSSIBLE = FW_FEATURE_LPAR | FW_FEATURE_PS3_LV1,
+	FW_FEATURE_PS3_ALWAYS = FW_FEATURE_LPAR | FW_FEATURE_PS3_LV1,
+	FW_FEATURE_NATIVE_POSSIBLE = 0,
+	FW_FEATURE_NATIVE_ALWAYS = 0,
 	FW_FEATURE_POSSIBLE =
 #ifdef CONFIG_PPC_PSERIES
 		FW_FEATURE_PSERIES_POSSIBLE |
@@ -70,6 +73,9 @@ enum {
 #ifdef CONFIG_PPC_PS3
 		FW_FEATURE_PS3_POSSIBLE |
 #endif
+#ifdef CONFIG_PPC_NATIVE
+		FW_FEATURE_NATIVE_ALWAYS |
+#endif
 		0,
 	FW_FEATURE_ALWAYS =
 #ifdef CONFIG_PPC_PSERIES
@@ -81,6 +87,9 @@ enum {
 #ifdef CONFIG_PPC_PS3
 		FW_FEATURE_PS3_ALWAYS &
 #endif
+#ifdef CONFIG_PPC_NATIVE
+		FW_FEATURE_NATIVE_ALWAYS &
+#endif
 		FW_FEATURE_POSSIBLE,
 
 #else /* CONFIG_PPC64 */
arch/powerpc/Kconfig +21 -5
Index: linux-2.6/arch/powerpc/Kconfig
===================================================================
@@ -390,6 +390,7 @@ config PPC_PSERIES
 	select PPC_RTAS
 	select RTAS_ERROR_LOGGING
 	select PPC_UDBG_16550
+	select PPC_NATIVE
 	default y
 
 config PPC_ISERIES
@@ -406,6 +407,7 @@ config PPC_CHRP
 	select PPC_RTAS
 	select PPC_MPC106
 	select PPC_UDBG_16550
+	select PPC_NATIVE
 	default y
 
 config PPC_EFIKA
@@ -414,6 +416,7 @@ config PPC_EFIKA
 	select PPC_RTAS
 	select RTAS_PROC
 	select PPC_MPC52xx
+	select PPC_NATIVE
 	default y
 
 config PPC_PMAC
@@ -422,6 +425,7 @@ config PPC_PMAC
 	select MPIC
 	select PPC_INDIRECT_PCI if PPC32
 	select PPC_MPC106 if PPC32
+	select PPC_NATIVE
 	default y
 
 config PPC_PMAC64
@@ -441,6 +445,7 @@ config PPC_PREP
 	select PPC_I8259
 	select PPC_INDIRECT_PCI
 	select PPC_UDBG_16550
+	select PPC_NATIVE
 	default y
 
 config PPC_MAPLE
@@ -452,6 +457,7 @@ config PPC_MAPLE
 	select GENERIC_TBSYNC
 	select PPC_UDBG_16550
 	select PPC_970_NAP
+	select PPC_NATIVE
 	default n
 	help
           This option enables support for the Maple 970FX Evaluation Board.
@@ -464,6 +470,7 @@ config PPC_PASEMI
 	select MPIC
 	select PPC_UDBG_16550
 	select GENERIC_TBSYNC
+	select PPC_NATIVE
 	help
 	  This option enables support for PA Semi's PWRficient line
 	  of SoC processors, including PA6T-1682M
@@ -478,6 +485,7 @@ config PPC_CELL_NATIVE
 	select PPC_DCR_MMIO
 	select PPC_OF_PLATFORM_PCI
 	select PPC_INDIRECT_IO
+	select PPC_NATIVE
 	select MPIC
 	default n
 
@@ -490,11 +498,6 @@ config PPC_IBM_CELL_BLADE
 	select PPC_UDBG_16550
 	select UDBG_RTAS_CONSOLE
 
-config UDBG_RTAS_CONSOLE
-	bool "RTAS based debug console"
-	depends on PPC_RTAS
-	default n
-
 config PPC_PS3
 	bool "Sony PS3"
 	depends on PPC_MULTIPLATFORM && PPC64
@@ -503,6 +506,19 @@ config PPC_PS3
 	  This option enables support for the Sony PS3 game console
 	  and other platforms using the PS3 hypervisor.
 
+config PPC_NATIVE
+	bool
+	depends on PPC_MULTIPLATFORM
+	help
+	  Support for running natively on the hardware, i.e. without
+	  a hypervisor. This option is not user-selectable but should
+	  be selected by all platforms that need it.
+
+config UDBG_RTAS_CONSOLE
+	bool "RTAS based debug console"
+	depends on PPC_RTAS
+	default n
+
 config XICS
 	depends on PPC_PSERIES
 	bool
arch/powerpc/platforms/ps3/mm.c +4 -0
Index: linux-2.6/arch/powerpc/platforms/ps3/mm.c
===================================================================
@@ -22,6 +22,7 @@
 #include <linux/module.h>
 #include <linux/memory_hotplug.h>
 
+#include <asm/firmware.h>
 #include <asm/lmb.h>
 #include <asm/udbg.h>
 #include <asm/ps3.h>
@@ -293,6 +294,9 @@ static int __init ps3_mm_add_memory(void
 	unsigned long start_pfn;
 	unsigned long nr_pages;
 
+	if (!firmware_has_feature(FW_FEATURE_PS3_LV1))
+		return 0;
+
 	BUG_ON(!mem_init_done);
 
 	start_addr = USE_LPAR_ADDR ? map.r1.base : map.rm.size;
arch/powerpc/platforms/ps3/setup.c +1 -1
Index: linux-2.6/arch/powerpc/platforms/ps3/setup.c
===================================================================
@@ -108,7 +108,7 @@ static int __init ps3_probe(void)
 	if (!of_flat_dt_is_compatible(dt_root, "PS3"))
 		return 0;
 
-	powerpc_firmware_features |= FW_FEATURE_LPAR;
+	powerpc_firmware_features |= FW_FEATURE_PS3_POSSIBLE;
 
 	ps3_os_area_init();
 	ps3_mm_init();
drivers/ps3/system-bus.c +4 -0
Index: linux-2.6/drivers/ps3/system-bus.c
===================================================================
@@ -27,6 +27,7 @@
 #include <asm/udbg.h>
 #include <asm/ps3.h>
 #include <asm/lv1call.h>
+#include <asm/firmware.h>
 
 #define dump_mmio_region(_a) _dump_mmio_region(_a, __func__, __LINE__)
 static void _dump_mmio_region(const struct ps3_mmio_region* r,
@@ -167,6 +168,9 @@ int __init ps3_system_bus_init(void)
 {
 	int result;
 
+	if (!firmware_has_feature(FW_FEATURE_PS3_LV1))
+		return 0;
+
 	result = bus_register(&ps3_system_bus_type);
 	BUG_ON(result);
 	return result;
arch/powerpc/oprofile/Makefile +1 -1
Index: linux-2.6/arch/powerpc/oprofile/Makefile
===================================================================
@@ -11,7 +11,7 @@ DRIVER_OBJS := $(addprefix ../../../driv
 		timer_int.o )
 
 oprofile-y := $(DRIVER_OBJS) common.o backtrace.o
-oprofile-$(CONFIG_PPC_CELL) += op_model_cell.o
+oprofile-$(CONFIG_PPC_CELL_NATIVE) += op_model_cell.o
 oprofile-$(CONFIG_PPC64) += op_model_rs64.o op_model_power4.o
 oprofile-$(CONFIG_FSL_BOOKE) += op_model_fsl_booke.o
 oprofile-$(CONFIG_6xx) += op_model_7450.o
arch/powerpc/oprofile/common.c +1 -1
Index: linux-2.6/arch/powerpc/oprofile/common.c
===================================================================
@@ -147,7 +147,7 @@ int __init oprofile_arch_init(struct opr
 
 	switch (cur_cpu_spec->oprofile_type) {
 #ifdef CONFIG_PPC64
-#ifdef CONFIG_PPC_CELL
+#ifdef CONFIG_PPC_CELL_NATIVE
 		case PPC_OPROFILE_CELL:
 			model = &op_model_cell;
 			break;
arch/powerpc/mm/Makefile +1 -1
Index: linux-2.6/arch/powerpc/mm/Makefile
===================================================================
@@ -8,7 +8,7 @@ endif
 
 obj-y				:= fault.o mem.o lmb.o
 obj-$(CONFIG_PPC32)		+= init_32.o pgtable_32.o mmu_context_32.o
-hash-$(CONFIG_PPC_MULTIPLATFORM) := hash_native_64.o
+hash-$(CONFIG_PPC_NATIVE)	:= hash_native_64.o
 obj-$(CONFIG_PPC64)		+= init_64.o pgtable_64.o mmu_context_64.o \
 				   hash_utils_64.o hash_low_64.o tlb_64.o \
 				   slb_low.o slb.o stab.o mmap.o imalloc.o \

--

Acked-by on v1 6/9
#10 lore raw

Re: [Cbe-oss-dev] [POWERPC 6/9] ps3: multiplatform build fixes

Quoted 12 lines · Arnd Bergmann

A few code paths need to check whether or not they are running on the PS3's LV1 hypervisor before making hcalls. This introduces a new firmware feature bit for this, FW_FEATURE_PS3_LV1.

Now when both PS3 and IBM_CELL_BLADE are enabled, but not PSERIES, FW_FEATURE_PS3_LV1 and FW_FEATURE_LPAR get enabled at compile time, which is a bug. The same problem can also happen for (PPC_ISERIES && !PPC_PSERIES && PPC_SOMETHING_ELSE). In order to solve this, I introduce a new CONFIG_PPC_NATIVE option that is set when at least one platform is selected that can run without a hypervisor and then turns the firmware feature check into a run-time option.

Quoted 6 lines

The new cell oprofile support that was recently merged does not work on hypervisor based platforms like the PS3, therefore make it depend on PPC_CELL_NATIVE instead of PPC_CELL. This may change if we get oprofile support for PS3.

#11 lore raw

PATCH 7/9 ps3: add a default zImage target

In reply to Arnd Bergmann

It's currently not possible to build the default zImage target if PS3 is the only selected platform. This is a hack to fall back to building the pseries style zImage, so the build is successful. This will probably change in the future, if someone writes a PS3 specific boot wrapper.

arch/powerpc/boot/Makefile +1 -0
Index: linux-2.6/arch/powerpc/boot/Makefile
===================================================================
@@ -155,6 +155,7 @@ $(obj)/uImage: vmlinux $(wrapperbits)
 image-$(CONFIG_PPC_PSERIES)		+= zImage.pseries
 image-$(CONFIG_PPC_MAPLE)		+= zImage.pseries
 image-$(CONFIG_PPC_IBM_CELL_BLADE)	+= zImage.pseries
+image-$(CONFIG_PPC_PS3)			+= zImage.pseries
 image-$(CONFIG_PPC_CHRP)		+= zImage.chrp
 image-$(CONFIG_PPC_PMAC)		+= zImage.pmac
 image-$(CONFIG_DEFAULT_UIMAGE)		+= uImage

--

#12 lore raw

Re: [POWERPC 7/9] ps3: add a default zImage target

Quoted 13 lines · Arnd Bergmann

It's currently not possible to build the default zImage target if PS3 is the only selected platform. This is a hack to fall back to building the pseries style zImage, so the build is successful. This will probably change in the future, if someone writes a PS3 specific boot wrapper.

Index: linux-2.6/arch/powerpc/boot/Makefile
===================================================================

arch/powerpc/boot/Makefile $(obj)/uImage: vmlinux $(wrapperbits) · line 161

Quoted 7 lines
arch/powerpc/boot/Makefile +1 -0
@@ -155,6 +155,7 @@ $(obj)/uImage: vmlinux $(wrapperbits)
 image-$(CONFIG_PPC_PSERIES)		+= zImage.pseries
 image-$(CONFIG_PPC_MAPLE)		+= zImage.pseries
 image-$(CONFIG_PPC_IBM_CELL_BLADE)	+= zImage.pseries
+image-$(CONFIG_PPC_PS3)			+= zImage.pseries
 image-$(CONFIG_PPC_CHRP)		+= zImage.chrp
 image-$(CONFIG_PPC_PMAC)		+= zImage.pmac
 image-$(CONFIG_DEFAULT_UIMAGE)		+= uImage

A better solution here is to make image-$(CONFIG_PPC_PS3) a stripped vmlinux, which the current PS3 bootloader (2.6.16 kboot) will be able to load.

I'll set that up and send out a new patch.

I started work on that zImage bootwrapper, but I have other items I'm tring to finish up.

-Geoff

#13 lore raw

PATCH 8/9 fix missing #include in sys_ppc32.c

In reply to Arnd Bergmann

sys_mmap is declared in asm/syscalls.h

arch/powerpc/kernel/sys_ppc32.c +1 -0
Index: linux-2.6/arch/powerpc/kernel/sys_ppc32.c
===================================================================
@@ -51,6 +51,7 @@
 #include <asm/time.h>
 #include <asm/mmu_context.h>
 #include <asm/ppc-pci.h>
+#include <asm/syscalls.h>
 
 /* readdir & getdents */
 #define NAME_OFFSET(de) ((int) ((de)->d_name - (char __user *) (de)))

--

#14 lore raw

PATCH 9/9 fix building without PCI

In reply to Arnd Bergmann

At least the ide driver calls pcibus_to_node, which is not defined when CONFIG_PCI is disabled. This adds a nop function for the !PCI case.

include/asm-powerpc/topology.h +7 -0
index 9fe7894..50c0140 100644
@@ -32,7 +32,14 @@ static inline int node_to_first_cpu(int 
 int of_node_to_nid(struct device_node *device);
 
 struct pci_bus;
+#ifdef CONFIG_PCI
 extern int pcibus_to_node(struct pci_bus *bus);
+#else
+static inline int pcibus_to_node(struct pci_bus *bus)
+{
+	return -1;
+}
+#endif
 
 #define pcibus_to_cpumask(bus)	(pcibus_to_node(bus) == -1 ? \
 					CPU_MASK_ALL : \

--