powerpc: Add hibernation support for FSL BookE processors
PATCH v2 1/1 powerpc: Add hibernation support for FSL BookE processors
This is started as swsusp_32.S modifications, but the amount of #ifdefs made the whole file horribly unreadable, so let's put the support into its own separate file.
The code should be relatively easy to modify to support 44x BookEs as well, but since I don't have any 44x to test, let's confine the code to FSL BookE. (The only FSL-specific parts are 'flush_dcache_L1' and TLB invalidation code).
---
Sorry for the delayed response...
On Tue, Jan 12, 2010 at 01:34:08PM +1100, Benjamin Herrenschmidt wrote:
[...]
Quoted 4 lines
Here's a quick review. Looks good but two things:
- Please make it
swsusp_booke.c, 44x support is trivial and I don't
want to rename the file :-)
Done.
Quoted 2 lines
- Is there really an SDR1 register on FSL BookE ? It's supposed to be
the pointer to the hash table on server ...
Thanks, fixed.
Quoted 2 lines
- You probably should save/restore the TCR and ack pending crap DEC or
FIT interrupts in the TSR right before you kick the decrementer
Done.
Quoted 2 lines
- Nowadays, we still assume that the "loader" kernel is exactly the
same as the "loaded" kernel on resume ?
I'm pretty sure today we do rely on this, yes. This is not some generic code limitation though, it's just hard to test the case when loader != loaded. In most cases it will work fine since the loader kernel wouldn't differ a lot from the loaded kernel, so it'll setup the low level stuff the same way.
We may try to link the loader kernel to a different address (relocate it as in kdump case), and hope that it'll trigger all sort of problems so that we could fix them.
Though, the better test case would be to resume the hibernated kernel directly from the bootloader.
Thanks,
arch/powerpc/kernel/Makefile | 8 +- arch/powerpc/kernel/swsusp_booke.S | 200 ++++++++++++++++++++++++++++++++++++ 2 files changed, 206 insertions(+), 2 deletions(-) create mode 100644 arch/powerpc/kernel/swsusp_booke.S
arch/powerpc/kernel/Makefile +6 -2
@@ -57,8 +57,12 @@ obj-$(CONFIG_CRASH_DUMP) += crash_dump.o obj-$(CONFIG_E500) += idle_e500.o obj-$(CONFIG_6xx) += idle_6xx.o l2cr_6xx.o cpu_setup_6xx.o obj-$(CONFIG_TAU) += tau_6xx.o -obj-$(CONFIG_HIBERNATION) += swsusp.o suspend.o \ - swsusp_$(CONFIG_WORD_SIZE).o +obj-$(CONFIG_HIBERNATION) += swsusp.o suspend.o +ifeq ($(CONFIG_FSL_BOOKE),y) +obj-$(CONFIG_HIBERNATION) += swsusp_booke.o +else +obj-$(CONFIG_HIBERNATION) += swsusp_$(CONFIG_WORD_SIZE).o +endif obj64-$(CONFIG_HIBERNATION) += swsusp_asm64.o obj-$(CONFIG_MODULES) += module.o module_$(CONFIG_WORD_SIZE).o obj-$(CONFIG_44x) += cpu_setup_44x.o
arch/powerpc/kernel/swsusp_booke.S new file +200 -0
200 lines not on this page · raw message
Signature · 1 line
1.7.0
Quoted 42 lines · Anton Vorontsov
This is started as
swsusp_32.Smodifications, but the amount of #ifdefs made the whole file horribly unreadable, so let's put the support into its own separate file.The code should be relatively easy to modify to support 44x BookEs as well, but since I don't have any 44x to test, let's confine the code to FSL BookE. (The only FSL-specific parts are '
flush_dcache_L1' and TLB invalidation code).Signed-off-by Anton Vorontsov <[email protected]>---
Sorry for the delayed response...
On Tue, Jan 12, 2010 at 01:34:08PM +1100, Benjamin Herrenschmidt wrote:
[...]Quoted 4 lines
Here's a quick review. Looks good but two things:
- Please make it
swsusp_booke.c, 44x support is trivial and I don't
want to rename the file :-)Done.
Quoted 2 lines
- Is there really an SDR1 register on FSL BookE ? It's supposed to be
the pointer to the hash table on server ...Thanks, fixed.
Quoted 2 lines
- You probably should save/restore the TCR and ack pending crap DEC or
FIT interrupts in the TSR right before you kick the decrementerDone.
Quoted 2 lines
- Nowadays, we still assume that the "loader" kernel is exactly the
same as the "loaded" kernel on resume ?I'm pretty sure today we do rely on this, yes. This is not some generic code limitation though, it's just hard to test the case when loader != loaded. In most cases it will work fine since the loader kernel wouldn't differ a lot from the loaded kernel, so it'll setup the low level stuff the same way.
Quoted 6 lines
We may try to link the loader kernel to a different address (relocate it as in kdump case), and hope that it'll trigger all sort of problems so that we could fix them.
Though, the better test case would be to resume the hibernated kernel directly from the bootloader.
Kumar,
According to patchwork, this is now delegated to you. Do you have any objections to merge this?
Thanks!
Quoted 232 lines
arch/powerpc/kernel/Makefile | 8 +- arch/powerpc/kernel/swsusp_booke.S | 200 ++++++++++++++++++++++++++++++++++++ 2 files changed, 206 insertions(+), 2 deletions(-) create mode 100644 arch/powerpc/kernel/swsusp_booke.Sarch/powerpc/kernel/Makefile +6
-2@@ -57,8 +57,12 @@ obj-$(CONFIG_CRASH_DUMP) += crash_dump.o obj-$(CONFIG_E500) += idle_e500.o obj-$(CONFIG_6xx) += idle_6xx.o l2cr_6xx.o cpu_setup_6xx.o obj-$(CONFIG_TAU) += tau_6xx.o -obj-$(CONFIG_HIBERNATION) += swsusp.o suspend.o \ - swsusp_$(CONFIG_WORD_SIZE).o +obj-$(CONFIG_HIBERNATION) += swsusp.o suspend.o +ifeq ($(CONFIG_FSL_BOOKE),y) +obj-$(CONFIG_HIBERNATION) += swsusp_booke.o +else +obj-$(CONFIG_HIBERNATION) += swsusp_$(CONFIG_WORD_SIZE).o +endif obj64-$(CONFIG_HIBERNATION) += swsusp_asm64.o obj-$(CONFIG_MODULES) += module.o module_$(CONFIG_WORD_SIZE).o obj-$(CONFIG_44x) += cpu_setup_44x.oarch/powerpc/kernel/swsusp_booke.S new file +200
-0@@ -0,0 +1,200 @@ +/* + * Based on swsusp_32.S, modified for FSL BookE by + * Anton Vorontsov <[email protected]> + * Copyright (c) 2009-2010 MontaVista Software, LLC. + */ + +#include <linux/threads.h> +#include <asm/processor.h> +#include <asm/page.h> +#include <asm/cputable.h> +#include <asm/thread_info.h> +#include <asm/ppc_asm.h> +#include <asm/asm-offsets.h> +#include <asm/mmu.h> + +/* + * Structure for storing CPU registers on the save area. + */ +#define SL_SP 0 +#define SL_PC 4 +#define SL_MSR 8 +#define SL_TCR 0xc +#define SL_SPRG0 0x10 +#define SL_SPRG1 0x14 +#define SL_SPRG2 0x18 +#define SL_SPRG3 0x1c +#define SL_SPRG4 0x20 +#define SL_SPRG5 0x24 +#define SL_SPRG6 0x28 +#define SL_SPRG7 0x2c +#define SL_TBU 0x30 +#define SL_TBL 0x34 +#define SL_R2 0x38 +#define SL_CR 0x3c +#define SL_LR 0x40 +#define SL_R12 0x44 /* r12 to r31 */ +#define SL_SIZE (SL_R12 + 80) + + .section .data + .align 5 + +_GLOBAL(swsusp_save_area) + .space SL_SIZE + + + .section .text + .align 5 + +_GLOBAL(swsusp_arch_suspend) + lis r11,swsusp_save_area@h + ori r11,r11,swsusp_save_area@l + + mflr r0 + stw r0,SL_LR(r11) + mfcr r0 + stw r0,SL_CR(r11) + stw r1,SL_SP(r11) + stw r2,SL_R2(r11) + stmw r12,SL_R12(r11) + + /* Save MSR & TCR */ + mfmsr r4 + stw r4,SL_MSR(r11) + mfspr r4,SPRN_TCR + stw r4,SL_TCR(r11) + + /* Get a stable timebase and save it */ +1: mftbu r4 + stw r4,SL_TBU(r11) + mftb r5 + stw r5,SL_TBL(r11) + mftbu r3 + cmpw r3,r4 + bne 1b + + /* Save SPRGs */ + mfsprg r4,0 + stw r4,SL_SPRG0(r11) + mfsprg r4,1 + stw r4,SL_SPRG1(r11) + mfsprg r4,2 + stw r4,SL_SPRG2(r11) + mfsprg r4,3 + stw r4,SL_SPRG3(r11) + mfsprg r4,4 + stw r4,SL_SPRG4(r11) + mfsprg r4,5 + stw r4,SL_SPRG5(r11) + mfsprg r4,6 + stw r4,SL_SPRG6(r11) + mfsprg r4,7 + stw r4,SL_SPRG7(r11) + + /* Call the low level suspend stuff (we should probably have made + * a stackframe... + */ + bl swsusp_save + + /* Restore LR from the save area */ + lis r11,swsusp_save_area@h + ori r11,r11,swsusp_save_area@l + lwz r0,SL_LR(r11) + mtlr r0 + + blr + +_GLOBAL(swsusp_arch_resume) + sync + + /* Load ptr the list of pages to copy in r3 */ + lis r11,(restore_pblist)@h + ori r11,r11,restore_pblist@l + lwz r3,0(r11) + + /* Copy the pages. This is a very basic implementation, to + * be replaced by something more cache efficient */ +1: + li r0,256 + mtctr r0 + lwz r5,pbe_address(r3) /* source */ + lwz r6,pbe_orig_address(r3) /* destination */ +2: + lwz r8,0(r5) + lwz r9,4(r5) + lwz r10,8(r5) + lwz r11,12(r5) + addi r5,r5,16 + stw r8,0(r6) + stw r9,4(r6) + stw r10,8(r6) + stw r11,12(r6) + addi r6,r6,16 + bdnz 2b + lwz r3,pbe_next(r3) + cmpwi 0,r3,0 + bne 1b + + bl flush_dcache_L1 + bl flush_instruction_cache + + lis r11,swsusp_save_area@h + ori r11,r11,swsusp_save_area@l + + lwz r4,SL_SPRG0(r11) + mtsprg 0,r4 + lwz r4,SL_SPRG1(r11) + mtsprg 1,r4 + lwz r4,SL_SPRG2(r11) + mtsprg 2,r4 + lwz r4,SL_SPRG3(r11) + mtsprg 3,r4 + lwz r4,SL_SPRG4(r11) + mtsprg 4,r4 + lwz r4,SL_SPRG5(r11) + mtsprg 5,r4 + lwz r4,SL_SPRG6(r11) + mtsprg 6,r4 + lwz r4,SL_SPRG7(r11) + mtsprg 7,r4 + + /* Invalidate TLB0 & TLB1 */ + li r6,0x04 + tlbivax 0,r6 + TLBSYNC + li r6,0x0c + tlbivax 0,r6 + TLBSYNC + + /* restore the MSR */ + lwz r3,SL_MSR(r11) + + /* Restore TB */ + li r3,0 + mttbl r3 + lwz r3,SL_TBU(r11) + lwz r4,SL_TBL(r11) + mttbu r3 + mttbl r4 + + /* Restore TCR and clear any pending bits in TSR. */ + lwz r4,SL_TCR(r11) + mtspr SPRN_TCR,r4 + lis r4, (TSR_ENW | TSR_WIS | TSR_DIS | TSR_FIS)@h + mtspr SPRN_TSR,r4 + + /* Kick decrementer */ + li r0,1 + mtdec r0 + + /* Restore the callee-saved registers and return */ + lwz r0,SL_CR(r11) + mtcr r0 + lwz r2,SL_R2(r11) + lmw r12,SL_R12(r11) + lwz r1,SL_SP(r11) + lwz r0,SL_LR(r11) + mtlr r0 + + li r3,0 + blrSignature · 1 line
1.7.0
Signature · 3 lines
Anton Vorontsov email: [email protected] irc://irc.freenode.net/bd2
Quoted 47 lines · Anton Vorontsov
Quoted 46 lines · Anton Vorontsov
This is started as
swsusp_32.Smodifications, but the amount of #ifdefs made the whole file horribly unreadable, so let's put the support into its own separate file.The code should be relatively easy to modify to support 44x BookEs as well, but since I don't have any 44x to test, let's confine the code to FSL BookE. (The only FSL-specific parts are '
flush_dcache_L1' and TLB invalidation code).Signed-off-by Anton Vorontsov <[email protected]>---
Sorry for the delayed response...
On Tue, Jan 12, 2010 at 01:34:08PM +1100, Benjamin Herrenschmidt wrote:
[...]Quoted 4 lines
Here's a quick review. Looks good but two things:
- Please make it
swsusp_booke.c, 44x support is trivial and I don't want to rename the file :-)Done.
Quoted 2 lines
- Is there really an SDR1 register on FSL BookE ? It's supposed to be the pointer to the hash table on server ...
Thanks, fixed.
Quoted 2 lines
- You probably should save/restore the TCR and ack pending crap DEC or FIT interrupts in the TSR right before you kick the decrementer
Done.
Quoted 2 lines
- Nowadays, we still assume that the "loader" kernel is exactly the same as the "loaded" kernel on resume ?
I'm pretty sure today we do rely on this, yes. This is not some generic code limitation though, it's just hard to test the case when loader != loaded. In most cases it will work fine since the loader kernel wouldn't differ a lot from the loaded kernel, so it'll setup the low level stuff the same way.
We may try to link the loader kernel to a different address (relocate it as in kdump case), and hope that it'll trigger all sort of problems so that we could fix them.
Quoted 7 lines
Quoted 2 lines
Though, the better test case would be to resume the hibernated kernel directly from the bootloader.
Kumar,
According to patchwork, this is now delegated to you. Do you have any objections to merge this?
Would like Scott's Ack.
- k
On Thu, Apr 15, 2010 at 02:20:23PM -0500, Kumar Gala wrote:
[...]
Quoted 6 lines
Quoted 4 lines
Kumar,
According to patchwork, this is now delegated to you. Do you have any objections to merge this?
Would like Scott's Ack.
Cc'ing.
Quoted 7 lines · Kumar Gala
Quoted 4 lines · Anton Vorontsov
Kumar,
According to patchwork, this is now delegated to you. Do you have any objections to merge this?
Would like Scott's Ack.
I think we need to save IACn, DACn, DBCRn, PID0, and USPRG0.
Might want to also save TLB1 contents, and maybe things like HIDn, cache registers, etc. -- I don't think they're changeable post-boot currently, but it'd be good to avoid surprises if that were to change.
Otherwise ACK.
-Scott
Quoted 7 lines · Scott Wood
Quoted 6 lines · Kumar Gala
Quoted 4 lines · Anton Vorontsov
Kumar,
According to patchwork, this is now delegated to you. Do you have any objections to merge this?
Quoted 7 lines
Quoted 1 line
Would like Scott's Ack.
I think we need to save IACn, DACn, DBCRn, PID0, and USPRG0.
Might want to also save TLB1 contents, and maybe things like HIDn, cache registers, etc. -- I don't think they're changeable post-boot currently, but it'd be good to avoid surprises if that were to change.
Otherwise ACK.
Oh, should also use the timebase SPRs rather than mftb/mttb.
-Scott
PATCH v3 1/1 powerpc: Add hibernation support for FSL BookE processors
This is started as swsusp_32.S modifications, but the amount of #ifdefs made the whole file horribly unreadable, so let's put the support into its own separate file.
The code should be relatively easy to modify to support 44x BookEs as well, but since I don't have any 44x to test, let's confine the code to FSL BookE. (The only FSL-specific parts are 'flush_dcache_L1' and TLB invalidation code).
---
Quoted 12 lines · Scott Wood
Quoted 11 lines · Scott Wood
Quoted 7 lines · Kumar Gala
Quoted 4 lines · Anton Vorontsov
Kumar,
According to patchwork, this is now delegated to you. Do you have any objections to merge this?
Would like Scott's Ack.
I think we need to save IACn, DACn, DBCRn, PID0, and USPRG0.
Quoted 8 lines
Quoted 6 lines
Might want to also save TLB1 contents, and maybe things like HIDn, cache registers, etc. -- I don't think they're changeable post-boot currently, but it'd be good to avoid surprises if that were to change.
Otherwise ACK.
Oh, should also use the timebase SPRs rather than mftb/mttb.
Hm. That's for old toolchains that don't understand mftb/mttb synonyms? OK, that would be better then.
I still don't save TLB, HIDs etc. Let's keep it simple for now.
arch/powerpc/kernel/Makefile | 8 +- arch/powerpc/kernel/swsusp_booke.S | 200 ++++++++++++++++++++++++++++++++++++ 2 files changed, 206 insertions(+), 2 deletions(-) create mode 100644 arch/powerpc/kernel/swsusp_booke.S
arch/powerpc/kernel/Makefile +6 -2
@@ -57,8 +57,12 @@ obj-$(CONFIG_CRASH_DUMP) += crash_dump.o obj-$(CONFIG_E500) += idle_e500.o obj-$(CONFIG_6xx) += idle_6xx.o l2cr_6xx.o cpu_setup_6xx.o obj-$(CONFIG_TAU) += tau_6xx.o -obj-$(CONFIG_HIBERNATION) += swsusp.o suspend.o \ - swsusp_$(CONFIG_WORD_SIZE).o +obj-$(CONFIG_HIBERNATION) += swsusp.o suspend.o +ifeq ($(CONFIG_FSL_BOOKE),y) +obj-$(CONFIG_HIBERNATION) += swsusp_booke.o +else +obj-$(CONFIG_HIBERNATION) += swsusp_$(CONFIG_WORD_SIZE).o +endif obj64-$(CONFIG_HIBERNATION) += swsusp_asm64.o obj-$(CONFIG_MODULES) += module.o module_$(CONFIG_WORD_SIZE).o obj-$(CONFIG_44x) += cpu_setup_44x.o
arch/powerpc/kernel/swsusp_booke.S new file +200 -0
200 lines not on this page · raw message
Signature · 1 line
1.7.0.3
arch/powerpc/kernel/swsusp_booke.S line 167
Quoted 7 lines · Anton Vorontsov
arch/powerpc/kernel/swsusp_booke.S new file +7
-0+ /* Invalidate TLB0 & TLB1 */ + li r6,0x04 + tlbivax 0,r6 + TLBSYNC + li r6,0x0c + tlbivax 0,r6 + TLBSYNC
Is this needed? Shouldn't the boot process have already given us a sane TLB?
-Scott
PATCH v4 1/1 powerpc: Add hibernation support for FSL BookE processors
This is started as swsusp_32.S modifications, but the amount of #ifdefs made the whole file horribly unreadable, so let's put the support into its own separate file.
The code should be relatively easy to modify to support 44x BookEs as well, but since I don't have any 44x to test, let's confine the code to FSL BookE. (The only FSL-specific part so far is 'flush_dcache_L1'.)
---
Quoted 11 lines · Scott Wood
Review
Quoted 7 lines · Anton Vorontsov
(quoted diff) +7
-0+ /* Invalidate TLB0 & TLB1 */ + li r6,0x04 + tlbivax 0,r6 + TLBSYNC + li r6,0x0c + tlbivax 0,r6 + TLBSYNCIs this needed? Shouldn't the boot process have already given us a sane TLB?
Thanks for catching, it seems that it's just a left over from some debugging code and not actually needed. ..that reminded me the time I spent inserting BookE specific code into swsusp_32.S, and then debugging all that #ifdef mess...
This is tested on e500v2.
arch/powerpc/kernel/Makefile | 8 +- arch/powerpc/kernel/swsusp_booke.S | 192 ++++++++++++++++++++++++++++++++++++ 2 files changed, 198 insertions(+), 2 deletions(-) create mode 100644 arch/powerpc/kernel/swsusp_booke.S
arch/powerpc/kernel/Makefile +6 -2
@@ -57,8 +57,12 @@ obj-$(CONFIG_CRASH_DUMP) += crash_dump.o obj-$(CONFIG_E500) += idle_e500.o obj-$(CONFIG_6xx) += idle_6xx.o l2cr_6xx.o cpu_setup_6xx.o obj-$(CONFIG_TAU) += tau_6xx.o -obj-$(CONFIG_HIBERNATION) += swsusp.o suspend.o \ - swsusp_$(CONFIG_WORD_SIZE).o +obj-$(CONFIG_HIBERNATION) += swsusp.o suspend.o +ifeq ($(CONFIG_FSL_BOOKE),y) +obj-$(CONFIG_HIBERNATION) += swsusp_booke.o +else +obj-$(CONFIG_HIBERNATION) += swsusp_$(CONFIG_WORD_SIZE).o +endif obj64-$(CONFIG_HIBERNATION) += swsusp_asm64.o obj-$(CONFIG_MODULES) += module.o module_$(CONFIG_WORD_SIZE).o obj-$(CONFIG_44x) += cpu_setup_44x.o
arch/powerpc/kernel/swsusp_booke.S new file +192 -0
192 lines not on this page · raw message
Signature · 1 line
1.7.0.3
Quoted 1 line · Anton Vorontsov
This is started as
swsusp_32.Smodifications, but the amount of =
#ifdefs
Quoted 5 lines
made the whole file horribly unreadable, so let's put the support into its own separate file.
=20
The code should be relatively easy to modify to support 44x BookEs as well, but since I don't have any 44x to test, let's confine the code =
to
Quoted 17 lines
FSL BookE. (The only FSL-specific part so far is '
flush_dcache_L1'.) =20Signed-off-by Anton Vorontsov <[email protected]>---
=20Quoted 11 lines · Scott Wood
Review
Quoted 7 lines · Anton Vorontsov
(quoted diff) +7
-0+ /* Invalidate TLB0 & TLB1 */ + li r6,0x04 + tlbivax 0,r6 + TLBSYNC + li r6,0x0c + tlbivax 0,r6 + TLBSYNC=20
Is this needed? Shouldn't the boot process have already given us a sane TLB?
Quoted 10 lines
=20
Thanks for catching, it seems that it's just a left over from some debugging code and not actually needed. ..that reminded me the time I spent inserting BookE specific code intoswsusp_32.S, and then debugging all that #ifdef mess...
=20
This is tested on e500v2.
=20arch/powerpc/kernel/Makefile| 8 +-arch/powerpc/kernel/swsusp_booke.S| 192 =
++++++++++++++++++++++++++++++++++++
Quoted 2 lines
2 files changed, 198 insertions(+), 2 deletions(-)create mode 100644
arch/powerpc/kernel/swsusp_booke.S
Scott, you ok (ACK) this version or still need tweaks?
- k=
arch/powerpc/kernel/swsusp_booke.S line 165
Quoted 5 lines · Anton Vorontsov
arch/powerpc/kernel/swsusp_booke.S new file +5
-0+ /* restore the MSR */ + lwz r3,SL_MSR(r11) + + /* Restore TB */ + li r3,0
Missing mtmsr?
Otherwise ACK.
-Scott
PATCH v5 1/1 powerpc: Add hibernation support for FSL BookE processors
This is started as swsusp_32.S modifications, but the amount of #ifdefs made the whole file horribly unreadable, so let's put the support into its own separate file.
The code should be relatively easy to modify to support 44x BookEs as well, but since I don't have any 44x to test, let's confine the code to FSL BookE. (The only FSL-specific part so far is 'flush_dcache_L1'.)
---
Quoted 10 lines · Scott Wood
Review
Quoted 5 lines · Anton Vorontsov
(quoted diff) +5
-0+ /* restore the MSR */ + lwz r3,SL_MSR(r11) + + /* Restore TB */ + li r3,0Missing mtmsr?
Otherwise ACK.
Fixed, thanks!
arch/powerpc/kernel/Makefile | 8 +- arch/powerpc/kernel/swsusp_booke.S | 193 ++++++++++++++++++++++++++++++++++++ 2 files changed, 199 insertions(+), 2 deletions(-) create mode 100644 arch/powerpc/kernel/swsusp_booke.S
arch/powerpc/kernel/Makefile +6 -2
@@ -57,8 +57,12 @@ obj-$(CONFIG_CRASH_DUMP) += crash_dump.o obj-$(CONFIG_E500) += idle_e500.o obj-$(CONFIG_6xx) += idle_6xx.o l2cr_6xx.o cpu_setup_6xx.o obj-$(CONFIG_TAU) += tau_6xx.o -obj-$(CONFIG_HIBERNATION) += swsusp.o suspend.o \ - swsusp_$(CONFIG_WORD_SIZE).o +obj-$(CONFIG_HIBERNATION) += swsusp.o suspend.o +ifeq ($(CONFIG_FSL_BOOKE),y) +obj-$(CONFIG_HIBERNATION) += swsusp_booke.o +else +obj-$(CONFIG_HIBERNATION) += swsusp_$(CONFIG_WORD_SIZE).o +endif obj64-$(CONFIG_HIBERNATION) += swsusp_asm64.o obj-$(CONFIG_MODULES) += module.o module_$(CONFIG_WORD_SIZE).o obj-$(CONFIG_44x) += cpu_setup_44x.o
arch/powerpc/kernel/swsusp_booke.S new file +193 -0
193 lines not on this page · raw message
Signature · 1 line
1.7.0.5
Kumar Gala
AppliedQuoted 1 line · Anton Vorontsov
This is started as
swsusp_32.Smodifications, but the amount of =
#ifdefs
Quoted 5 lines
made the whole file horribly unreadable, so let's put the support into its own separate file.
=20
The code should be relatively easy to modify to support 44x BookEs as well, but since I don't have any 44x to test, let's confine the code =
to
Quoted 17 lines
FSL BookE. (The only FSL-specific part so far is '
flush_dcache_L1'.) =20Signed-off-by Anton Vorontsov <[email protected]>Acked-by Scott Wood <[email protected]>---
=20Quoted 10 lines · Scott Wood
arch/powerpc/kernel/swsusp_booke.S line 166
Quoted 5 lines · Anton Vorontsov
arch/powerpc/kernel/swsusp_booke.S new file +5
-0+ /* restore the MSR */ + lwz r3,SL_MSR(r11) + + /* Restore TB */ + li r3,0=20
Missing mtmsr?
=20
Otherwise ACK.
Quoted 5 lines
=20
Fixed, thanks!
=20arch/powerpc/kernel/Makefile| 8 +-arch/powerpc/kernel/swsusp_booke.S| 193 =
++++++++++++++++++++++++++++++++++++
Quoted 2 lines
2 files changed, 199 insertions(+), 2 deletions(-)create mode 100644
arch/powerpc/kernel/swsusp_booke.S
applied to next (fixed one white space issue).
- k=
Quoted 1 line · Anton Vorontsov
This is started as
swsusp_32.Smodifications, but the amount of =
#ifdefs
Quoted 5 lines
made the whole file horribly unreadable, so let's put the support into its own separate file.
=20
The code should be relatively easy to modify to support 44x BookEs as well, but since I don't have any 44x to test, let's confine the code =
to
Quoted 17 lines
FSL BookE. (The only FSL-specific part so far is '
flush_dcache_L1'.) =20Signed-off-by Anton Vorontsov <[email protected]>Acked-by Scott Wood <[email protected]>---
=20Quoted 10 lines · Scott Wood
arch/powerpc/kernel/swsusp_booke.S line 166
Quoted 5 lines · Anton Vorontsov
arch/powerpc/kernel/swsusp_booke.S new file +5
-0+ /* restore the MSR */ + lwz r3,SL_MSR(r11) + + /* Restore TB */ + li r3,0=20
Missing mtmsr?
=20
Otherwise ACK.
Quoted 5 lines
=20
Fixed, thanks!
=20arch/powerpc/kernel/Makefile| 8 +-arch/powerpc/kernel/swsusp_booke.S| 193 =
++++++++++++++++++++++++++++++++++++
Quoted 2 lines
2 files changed, 199 insertions(+), 2 deletions(-)create mode 100644
arch/powerpc/kernel/swsusp_booke.S
Is there board specific code to go along with this?
When I enable I get:
arch/powerpc/kernel/built-in.o: In function `cpu_idle':/home/galak/git/master/powerpc/arch/powerpc/kernel/idle.c:98: undefined = reference to `cpu_die'
make: *** [.tmp_vmlinux1] Error 1
- k=
On Mon, May 17, 2010 at 04:22:37PM -0500, Kumar Gala wrote:
[...]
Quoted 6 lines
Quoted 4 lines
arch/powerpc/kernel/Makefile| 8 +-arch/powerpc/kernel/swsusp_booke.S| 193 ++++++++++++++++++++++++++++++++++++2 files changed, 199 insertions(+), 2 deletions(-)create mode 100644
arch/powerpc/kernel/swsusp_booke.SIs there board specific code to go along with this?
Nope.
Quoted 5 lines
When I enable I get:
arch/powerpc/kernel/built-in.o: In function `cpu_idle':/home/galak/git/master/powerpc/arch/powerpc/kernel/idle.c:98: undefined reference to `cpu_die'
make: *** [.tmp_vmlinux1] Error 1
This isn't caused by hibernation or my patch specifically. Instead, it's a long standing issue, enabling CONFIG_SMP and CONFIG_SUSPEND is enough to trigger this. Patch exists:
http://www.mail-archive.com/[email protected]/msg27982.html
No idea why it didn't make Linus' tree.
Signature · 3 lines
Anton Vorontsov email: [email protected] irc://irc.freenode.net/bd2
Quoted 10 lines · Scott Wood
Quoted 7 lines · Kumar Gala
Quoted 4 lines · Anton Vorontsov
Kumar,
According to patchwork, this is now delegated to you. Do you have any objections to merge this?
Would like Scott's Ack.
I think we need to save IACn, DACn, DBCRn,
Does the kernel actually need these registers? I mean, they're saved per thread anyway.
Quoted 1 line
PID0,
Kernel clears it early at boot, why would we save it?
Quoted 1 line
and USPRG0.
Currently this isn't used at all.
Quoted 4 lines
Might want to also save TLB1 contents, and maybe things like HIDn, cache registers, etc. -- I don't think they're changeable post-boot currently, but it'd be good to avoid surprises if that were to change.
Hm. I don't really like the idea of doing things 'just in case', it just adds a code which functionality isn't tested, and which we will might never actually need.
Thanks,
Signature · 3 lines
Anton Vorontsov email: [email protected] irc://irc.freenode.net/bd2
Quoted 8 lines · Anton Vorontsov
Quoted 7 lines · Scott Wood
Quoted 6 lines · Kumar Gala
Quoted 4 lines · Anton Vorontsov
Kumar,
According to patchwork, this is now delegated to you. Do you have any objections to merge this?
Would like Scott's Ack.
Quoted 8 lines
Quoted 1 line
I think we need to save IACn, DACn, DBCRn,
Does the kernel actually need these registers? I mean, they're saved per thread anyway.
Quoted 1 line
PID0,
Kernel clears it early at boot, why would we save it?
What context are we in when we suspend, and what context is expected after resume? If we're guaranteed not to be on a thread that cares about any of this, then OK.
Quoted 3 lines
Quoted 1 line
and USPRG0.
Currently this isn't used at all.
Was thinking that it's up to userspace whether to use it or not -- though it ought to be saved on thread switch instead.
Quoted 8 lines
Quoted 4 lines
Might want to also save TLB1 contents, and maybe things like HIDn, cache registers, etc. -- I don't think they're changeable post-boot currently, but it'd be good to avoid surprises if that were to change.
Hm. I don't really like the idea of doing things 'just in case', it just adds a code which functionality isn't tested, and which we will might never actually need.
The intent was to reduce dependencies between the suspend code and what the rest of the kernel does -- it's likely that someone adding some runtime manipulation of one of these things (e.g. hugetlbfs modifying TLB1) would not think to check suspend/hibernation code. It would still be somewhat tested, in that you would see problems if you restored the wrong values (just not if you failed to restore at all).
We'll need to save/restore those things anyway for deep sleep -- but I'm not sure whether it'll be feasible to share code with hibernate (I recall trying on 83xx and giving up). In any case, it can probably wait until 85xx deep sleep gets pushed out.
-Scott
Quoted 14 lines · Scott Wood
Quoted 13 lines · Anton Vorontsov
Quoted 8 lines · Scott Wood
Quoted 6 lines · Kumar Gala
Quoted 4 lines · Anton Vorontsov
Kumar,
According to patchwork, this is now delegated to you. Do you have any objections to merge this?
Would like Scott's Ack.
I think we need to save IACn, DACn, DBCRn,
Does the kernel actually need these registers? I mean, they're saved per thread anyway.
Quoted 7 lines
Quoted 3 lines
Quoted 1 line
PID0,
Kernel clears it early at boot, why would we save it?
What context are we in when we suspend, and what context is expected after resume? If we're guaranteed not to be on a thread that cares about any of this, then OK.
Yeah, we're in the kernel context on suspend and resume.
Quoted 16 lines
Quoted 3 lines
Quoted 1 line
and USPRG0.
Currently this isn't used at all.
Was thinking that it's up to userspace whether to use it or not -- though it ought to be saved on thread switch instead.
Quoted 8 lines
Quoted 4 lines
Might want to also save TLB1 contents, and maybe things like HIDn, cache registers, etc. -- I don't think they're changeable post-boot currently, but it'd be good to avoid surprises if that were to change.
Hm. I don't really like the idea of doing things 'just in case', it just adds a code which functionality isn't tested, and which we will might never actually need.
Quoted 7 lines
The intent was to reduce dependencies between the suspend code and what the rest of the kernel does -- it's likely that someone adding some runtime manipulation of one of these things (e.g. hugetlbfs modifying TLB1) would not think to check suspend/hibernation code. It would still be somewhat tested, in that you would see problems if you restored the wrong values (just not if you failed to restore at all).
Yep, I understand the intent, though, I'd like 'git reset --hard this_commit' to bring a minimal, but still fully functional hibernate, without unneeded cruft. We might add more savings/restores later [or sooner, when needed], but that won't matter that much since we'll have a point in the git history with that minimal support.
I could even implement more savings and restores on top of this patch, but really, I don't see a hurry in doing this now. :-) So that's why I don't.
Thanks,
Signature · 3 lines
Anton Vorontsov email: [email protected] irc://irc.freenode.net/bd2