Skip to contentSkip to conversation

What's going on here ?

DISCUSSION Benjamin Herrenschmidt 3 participants 4 messages Mar 15, 2005

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

Hi just see that the whole stack of pmac SWSUSP patches went in, without any notice nor CC nor anything from any of the PPC maintainers ! That is a bit annoying don't you think ?

Paulus and I wrote most of those patches, granted, and they've been hanging around for some time, but I had good reasons not to submit them in their current state.

And regardless, I'm pretty pissed off by the fact that such invasive changes to the architecture and the platform support were submitted and merged without any notice nor ack from any of the arch or platform maintainers (basically paulus and me).

Ben.

#1 lore raw

Hi!

Quoted 4 lines

Hi just see that the whole stack of pmac SWSUSP patches went in, without any notice nor CC nor anything from any of the PPC maintainers ! That is a bit annoying don't you think ?

Quoted 8 lines

Paulus and I wrote most of those patches, granted, and they've been hanging around for some time, but I had good reasons not to submit them in their current state.

And regardless, I'm pretty pissed off by the fact that such invasive changes to the architecture and the platform support were submitted and merged without any notice nor ack from any of the arch or platform maintainers (basically paulus and me).

Sorry, that's probably my fault.

rjw and hugang did (pretty neccessary) changes to base swsusp (pagedir table -> pagedir linklist), that unfortunately needed update to all the assembly parts. It was series 1/3 update core, i386 and x86-64, 2/3 update ppc, 3/3 introduce initramfs.

This is the offending patch I believe (but the version that was merged was From: me, without code changes).

I realized that patch does more than changing from table to linklist, but it looked mostly okay, so I forwarded it. Sorry.

So, what to do now?

a) just revert it

or

b) revert pmac_setup.c and via-pmu parts and Kconfig part

or

c) just disable Kconfig part and fix it up with incremental patches

?
Pavel

From: "Rafael J. Wysocki" <[email protected]>
To: Andrew Morton <[email protected]>
Subject: [PATCH][3/3] swsusp: use non-contiguous memory
Cc: Hu Gang <[email protected]>,
LKML <[email protected]>, Pavel Machek <[email protected]>
Lines: 686

From: Hu Gang <[email protected]>

Subject: swsusp: use non-contiguous memory on resume - ppc support

This patch contains the architecture-dependent changes for ppc required for using a linklist instead of an array of page backup entries during resume.

arch/ppc/Kconfig +2 -0
diff -Nru linux-2.6.11-a/arch/ppc/Kconfig linux-2.6.11-b/arch/ppc/Kconfig
@@ -1046,6 +1046,8 @@
 
 source "drivers/zorro/Kconfig"
 
+source kernel/power/Kconfig
+
 endmenu
 
 menu "Bus options"
arch/ppc/kernel/asm-offsets.c +5 -0
diff -Nru linux-2.6.11-a/arch/ppc/kernel/asm-offsets.c linux-2.6.11-b/arch/ppc/kernel/asm-offsets.c
@@ -16,6 +16,7 @@
 #include <linux/string.h>
 #include <linux/types.h>
 #include <linux/ptrace.h>
+#include <linux/suspend.h>
 #include <linux/mman.h>
 #include <linux/mm.h>
 #include <asm/io.h>
@@ -136,6 +137,10 @@
 	DEFINE(TI_CPU, offsetof(struct thread_info, cpu));
 	DEFINE(TI_PREEMPT, offsetof(struct thread_info, preempt_count));
 
+	DEFINE(pbe_address, offsetof(struct pbe, address));
+	DEFINE(pbe_orig_address, offsetof(struct pbe, orig_address));
+	DEFINE(pbe_next, offsetof(struct pbe, next));
+
 	DEFINE(NUM_USER_SEGMENTS, TASK_SIZE>>28);
 	return 0;
 }
arch/ppc/kernel/Makefile +1 -0
diff -Nru linux-2.6.11-a/arch/ppc/kernel/Makefile linux-2.6.11-b/arch/ppc/kernel/Makefile
@@ -16,6 +16,7 @@
 					semaphore.o syscalls.o setup.o \
 					cputable.o ppc_htab.o perfmon.o
 obj-$(CONFIG_6xx)		+= l2cr.o cpu_setup_6xx.o
+obj-$(CONFIG_SOFTWARE_SUSPEND)	+= swsusp.o
 obj-$(CONFIG_POWER4)		+= cpu_setup_power4.o
 obj-$(CONFIG_MODULES)		+= module.o ppc_ksyms.o
 obj-$(CONFIG_NOT_COHERENT_CACHE)	+= dma-mapping.o
arch/ppc/kernel/signal.c +10 -0
diff -Nru linux-2.6.11-a/arch/ppc/kernel/signal.c linux-2.6.11-b/arch/ppc/kernel/signal.c
@@ -28,6 +28,7 @@
 #include <linux/elf.h>
 #include <linux/tty.h>
 #include <linux/binfmts.h>
+#include <linux/suspend.h>
 #include <asm/ucontext.h>
 #include <asm/uaccess.h>
 #include <asm/pgtable.h>
@@ -704,6 +705,14 @@
 	unsigned long frame, newsp;
 	int signr, ret;
 
+	if (current->flags & PF_FREEZE) {
+		refrigerator(PF_FREEZE);
+		signr = 0;
+		ret = regs->gpr[3];
+		if (!signal_pending(current))
+			goto no_signal;
+	}
+
 	if (!oldset)
 		oldset = &current->blocked;
 
@@ -726,6 +735,7 @@
 			regs->gpr[3] = EINTR;
 			/* note that the cr0.SO bit is already set */
 		} else {
+no_signal:
 			regs->nip -= 4;	/* Back up & retry system call */
 			regs->result = 0;
 			regs->trap = 0;
arch/ppc/kernel/swsusp.S +349 -0

349 lines not on this page · raw message

arch/ppc/kernel/vmlinux.lds.S +6 -0
diff -Nru linux-2.6.11-a/arch/ppc/kernel/vmlinux.lds.S linux-2.6.11-b/arch/ppc/kernel/vmlinux.lds.S
@@ -74,6 +74,12 @@
     CONSTRUCTORS
   }
 
+  . = ALIGN(4096);
+  __nosave_begin = .;
+  .data_nosave : { *(.data.nosave) }
+  . = ALIGN(4096);
+  __nosave_end = .;
+
   . = ALIGN(32);
   .data.cacheline_aligned : { *(.data.cacheline_aligned) }
 
arch/ppc/platforms/pmac_setup.c +55 -0

77 lines not on this page · raw message

drivers/macintosh/via-pmu.c +84 -2

110 lines not on this page · raw message

include/asm-ppc/suspend.h +12 -0
diff -Nru linux-2.6.11-a/include/asm-ppc/suspend.h linux-2.6.11-b/include/asm-ppc/suspend.h
@@ -0,0 +1,12 @@
+static inline int arch_prepare_suspend(void)
+{
+	return 0;
+}
+
+static inline void save_processor_state(void)
+{
+}
+
+static inline void restore_processor_state(void)
+{
+}
include/linux/suspend.h +1 -1
diff -Nru linux-2.6.11-a/include/linux/suspend.h linux-2.6.11-b/include/linux/suspend.h
@@ -1,7 +1,7 @@
 #ifndef _LINUX_SWSUSP_H
 #define _LINUX_SWSUSP_H
 
-#if defined(CONFIG_X86) || defined(CONFIG_FRV)
+#if defined(CONFIG_X86) || defined(CONFIG_FRV) || defined(CONFIG_PPC32)
 #include <asm/suspend.h>
 #endif
 #include <linux/swap.h>
Signature · 2 lines
People were complaining that M$ turns users into beta-testers...
...jr ghea gurz vagb qrirybcref, naq gurl frrz gb yvxr vg gung jnl!
#2 lore raw
Quoted 11 lines

rjw and hugang did (pretty neccessary) changes to base swsusp (pagedir table -> pagedir linklist), that unfortunately needed update to all the assembly parts. It was series 1/3 update core, i386 and x86-64, 2/3 update ppc, 3/3 introduce initramfs.

This is the offending patch I believe (but the version that was merged was From: me, without code changes).

I realized that patch does more than changing from table to linklist, but it looked mostly okay, so I forwarded it. Sorry.

It does more than that ... it _adds_ swsusp to ppc ! swsusp wasn't in mainline at all for ppc because I consider it not ready. And even the asm change should go through me anyway since i wrote that code and I'm not sure they know all the possible "issues" with that code.

Quoted 11 lines

So, what to do now?

a) just revert it

or

b) revert pmac_setup.c and via-pmu parts and Kconfig part

or

c) just disable Kconfig part and fix it up with incremental patches

I'll decide later today. I may well keep it and do the cleanup I had in mind on top of this, which means merging the pmac suspend-to-ram with the common infrastructure. But that will need some changes & hooks to the core swsusp.

Ben.

#3 lore raw
Quoted 9 lines · Benjamin Herrenschmidt
Quoted 9 lines

rjw and hugang did (pretty neccessary) changes to base swsusp (pagedir table -> pagedir linklist), that unfortunately needed update to all the assembly parts. It was series 1/3 update core, i386 and x86-64, 2/3 update ppc, 3/3 introduce initramfs.

This is the offending patch I believe (but the version that was merged was From: me, without code changes).

Quoted 19 lines
Quoted 2 lines

I realized that patch does more than changing from table to linklist, but it looked mostly okay, so I forwarded it. Sorry.

It does more than that ... it _adds_ swsusp to ppc ! swsusp wasn't in mainline at all for ppc because I consider it not ready. And even the asm change should go through me anyway since i wrote that code and I'm not sure they know all the possible "issues" with that code.

Quoted 11 lines

So, what to do now?

a) just revert it

or

b) revert pmac_setup.c and via-pmu parts and Kconfig part

or

c) just disable Kconfig part and fix it up with incremental patches

I hope that's can merge into, It works fine in my PowerBook G4.

Quoted 7 lines

I'll decide later today. I may well keep it and do the cleanup I had in mind on top of this, which means merging the pmac suspend-to-ram with the common infrastructure. But that will need some changes & hooks to the core swsusp.

Ben.

Signature · 5 lines
Hu Gang       .-.
              /v\
             // \\ 
Linux User  /(   )\  [204016]
GPG Key ID   ^^-^^   http://soulinfo.com/~hugang/hugang.asc