gmane.linux.power-management.general

Re: resume from ram hangs on 2.6.26+ kernels on thinkpadz60m

Rafael J. Wysocki — 2008-08-21T12:43:43

Not really. We must have broken something in 2.6.26, but there were not too many suspend-specific patches in there vs .25. What s2ram options are used? What hardware is there in the box (chipset, graphics)? Thanks, Rafael

Re: forcedeth 10de:0373 doesn't work on resume

Rafael J. Wysocki — 2008-08-21T12:34:34

I cannot reproduce the problem locally, at least not with 2.6.27-rc4. I was able to make my box resume from suspend to RAM by using # s2ram --force --vbe_mode --vbe_post (s2ram as shipped in openSUSE 11.0) and forcedeth works correctly after the resume. However, I had to add 'acpi_sleep=old_ordering' to the kernel command line. My box is a desktop with an Asus A8N-SLI motherboard and an Athlon 64 X2 CPU. I have created a Bugzilla entry for this issue at http://bugzilla.kernel.org/show_bug.cgi?id=11390 Please put a boot log from dmesg and the output of 'lspci -vvv' in there, as text attachments. Thanks, Rafael

Re: [PATCH 2/6] Container Freezer: uninlinethaw_process()

Pavel Machek — 2008-08-20T16:44:36

20 bytes... Does it still compile out if freezer is disabled? Pavel

Re: [PATCH 1/6] Container Freezer: Fix freezer Kconfig

Pavel Machek — 2008-08-20T16:42:12

Acked-by: Pavel Machek suse.cz>

Re: Power management for individual devices

Pavel Machek — 2008-08-20T16:18:15

Hi! Just suspend devices when they are not used; you should not need new interface(s) to do that.

[PATCH 6/6] Container Freezer: Document the cgroupfreezer subsystem.

Matt Helsley — 2008-08-19T23:22:38

Describe why we need the freezer subsystem and how to use it in a documentation file. Since the cgroups.txt file is focused on the subsystem-agnostic portions of cgroups make a directory and move the old cgroups.txt file at the same time. Signed-off-by: Matt Helsley us.ibm.com> Cc: Andrew Morton linux-foundation.org> Cc: Paul Menage google.com> Cc: containers< at >lists.linux-foundation.org --- Documentation/cgroups.txt | 548 ---------------------------- Documentation/cgroups/cgroups.txt | 548 ++++++++++++++++++++++++++++ Documentation/cgroups/freezer-subsystem.txt | 99 +++++ Documentation/cpusets.txt | 2 4 files changed, 648 insertions(+), 549 deletions(-) Index: linux-2.6.27-rc1-mm1/Documentation/cgroups.txt =================================================================== --- linux-2.6.27-rc1-mm1.orig/Documentation/cgroups.txt +++ /dev/null < at >< at > -1,548 +0,0 < at >< at > -CGROUPS -------- - -Written by Paul Menage google.

[PATCH 3/6] Container Freezer: Make freezer state namesless generic

Matt Helsley — 2008-08-19T23:22:35

Rename cgroup freezer states to be less generic to avoid any name collisions while also better describing what each state is. Signed-off-by: Matt Helsley us.ibm.com> Cc: Andrew Morton linux-foundation.org> --- kernel/cgroup_freezer.c | 56 ++++++++++++++++++++++++------------------------ 1 file changed, 28 insertions(+), 28 deletions(-) Index: linux-2.6.27-rc1-mm1/kernel/cgroup_freezer.c =================================================================== --- linux-2.6.27-rc1-mm1.orig/kernel/cgroup_freezer.c +++ linux-2.6.27-rc1-mm1/kernel/cgroup_freezer.c < at >< at > -22,9 +22,9 < at >< at > #include enum freezer_state { -STATE_RUNNING = 0, -STATE_FREEZING, -STATE_FROZEN, +CGROUP_THAWED = 0, +CGROUP_FREEZING, +CGROUP_FROZEN, }; struct freezer { < at >< at > -57,7 +57,7 < at >< at > int cgroup_frozen(struct task_struct *ta state = freezer->state; task_unlock(task); -return state == STATE_FROZEN; +return state == CGROUP_FROZEN; } /* < at >< at > -65,7 +65,7 < at >< at > int cgroup_frozen(struct task_str

[PATCH 5/6] Container Freezer: Rename check_if_frozen()

Matt Helsley — 2008-08-19T23:22:37

check_if_frozen() sounds like it should return something when in fact it's just updating the freezer state. Signed-off-by: Matt Helsley us.ibm.com> Cc: Andrew Morton linux-foundation.org> --- kernel/cgroup_freezer.c | 12 ++++++++---- 1 file changed, 8 insertions(+), 4 deletions(-) Index: linux-2.6.27-rc1-mm1/kernel/cgroup_freezer.c =================================================================== --- linux-2.6.27-rc1-mm1.orig/kernel/cgroup_freezer.c +++ linux-2.6.27-rc1-mm1/kernel/cgroup_freezer.c < at >< at > -201,8 +201,8 < at >< at > static void freezer_fork(struct cgroup_s /* * caller must hold freezer->lock */ -static void check_if_frozen(struct cgroup *cgroup, - struct freezer *freezer) +static void update_freezer_state(struct cgroup *cgroup, + struct freezer *freezer) { struct cgroup_iter it; struct task_struct *task; < at >< at > -222,6 +222,10 < at >< at > static void check_if_frozen(struct cgrou */ if (nfrozen == ntotal) freezer->state = CGROUP_FROZEN; +else if (nfrozen > 0) +freeze

[PATCH 0/6] Container Freezer: fixes for -mm

Matt Helsley — 2008-08-19T23:22:32

Hi Andrew, These patches stack on top of the series you already have for the container freezer patches. I believe they address all of the comments from you and Rafael so far. Please consider these for -mm. [PATCH 1/6] Container Freezer: Fix freezer Kconfig [PATCH 2/6] Container Freezer: uninline thaw_process() [PATCH 3/6] Container Freezer: Make freezer state names less generic [PATCH 4/6] Container Freezer: Cleanup comment [PATCH 5/6] Container Freezer: Rename check_if_frozen() [PATCH 6/6] Container Freezer: Document the cgroup freezer subsystem Cheers, -Matt Helsley

[PATCH 4/6] Container Freezer: Cleanup comment.

Matt Helsley — 2008-08-19T23:22:36

Make this comment clearer. Signed-off-by: Matt Helsley us.ibm.com> Cc: Andrew Morton linux-foundation.org> --- Andrew was wondering if the comment was in the right place. I think it is but suspect it could stand to be clarified. kernel/cgroup_freezer.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) Index: linux-2.6.27-rc1-mm1/kernel/cgroup_freezer.c =================================================================== --- linux-2.6.27-rc1-mm1.orig/kernel/cgroup_freezer.c +++ linux-2.6.27-rc1-mm1/kernel/cgroup_freezer.c < at >< at > -61,8 +61,8 < at >< at > int cgroup_frozen(struct task_struct *ta } /* - * Buffer size for freezer state is limited by cgroups write_string() - * interface. See cgroups code for the current size. + * cgroups_write_string() limits the size of freezer state strings to + * CGROUP_LOCAL_BUFFER_SIZE */ static const char *freezer_state_strs[] = { "THAWED",

[PATCH 2/6] Container Freezer: uninline thaw_process()

Matt Helsley — 2008-08-19T23:22:34

Now that the cgroup freezer system also calls thaw_process() inlining these functions uses more space. Uninlining returns some space: Before: textdatabssdechexfilename 42608722755322908164827220 49a854vmlinux After: textdatabssdechexfilename 4260852275532290816482720049a840vmlinux Signed-off-by: Matt Helsley us.ibm.com> Cc: Andrew Morton linux-foundation.org> Cc: Rafael J. Wysocki sisk.pl> --- include/linux/freezer.h |29 +++-------------------------- kernel/freezer.c |31 +++++++++++++++++++++++++++++++ 2 files changed, 34 insertions(+), 26 deletions(-) Index: linux-2.6.27-rc1-mm1/include/linux/freezer.h =================================================================== --- linux-2.6.27-rc1-mm1.orig/include/linux/freezer.h +++ linux-2.6.27-rc1-mm1/include/linux/freezer.h < at >< at > -46,34 +46,11 < at >< at > static inline bool should_send_signal(st /* * Wake up a frozen process - * - * task_lock() is needed to prevent the race with refrigerator() which may - * oc

[PATCH 1/6] Container Freezer: Fix freezer Kconfig

Matt Helsley — 2008-08-19T23:22:33

kernel/power/Kconfig is not sourced from all architectures but the freezer code should be available to all architectures for the cgroup freezer subsystem. Sourcing a new kernel/Kconfig.freezer has the added advantage of keeping the config definition close to its use in the kernel/Makefile. Signed-off-by: Matt Helsley us.ibm.com> Cc: Andrew Morton linux-foundation.org> Cc: Rafael J. Wysocki sisk.pl> --- This patch should probably be merged with the patch that put: config FREEZER in kernel/power/Kconfig arch/alpha/Kconfig | 1 + arch/arm/Kconfig | 2 ++ arch/avr32/Kconfig | 2 ++ arch/blackfin/Kconfig | 3 +++ arch/cris/Kconfig | 2 ++ arch/frv/Kconfig | 2 ++ arch/h8300/Kconfig | 2 ++ arch/ia64/Kconfig | 2 ++ arch/m32r/Kconfig | 2 ++ arch/m68k/Kconfig | 2 ++ arch/m68knommu/Kconfig | 2 ++ arch/mips/Kconfig | 2 ++ arch/mn10300/Kconfig | 2 ++ arch/parisc/Kconfig | 2 ++ arch/powerpc/Kco

Re: [PATCH] PCI PM: Introduce function pci_wake_from_d3(rev. 2)

Pavel Machek — 2008-08-19T09:29:57

Hi! ACK. Pavel

Re: forcedeth 10de:0373 doesn't work on resume

Rafael J. Wysocki — 2008-08-18T21:48:52

This was a question for me. :-)

Re: forcedeth 10de:0373 doesn't work on resume

Simon Arlott — 2008-08-18T21:43:02

I'm not sure what you're asking here...?

Re: forcedeth 10de:0373 doesn't work on resume

Rafael J. Wysocki — 2008-08-18T21:42:31

Well, maybe. The only way to verify that would be to try with MSI disabled. ;-) Thanks, Rafael

Re: forcedeth 10de:0373 doesn't work on resume

Rafael J. Wysocki — 2008-08-18T21:39:39

Yes. Usually the BIOS does something to devices in that case. OK, I will. That's interesting. I'll try to reproduce it. Hm. I wonder if that's specific to forcedeth or other drivers may be affected. Thanks, Rafael

Re: forcedeth 10de:0373 doesn't work on resume

Yinghai Lu — 2008-08-18T21:34:58

same problem with nv_shutdown about setting it to D3Hot? YH

Re: forcedeth 10de:0373 doesn't work on resume

Simon Arlott — 2008-08-18T21:25:07

With hibernation would it re-initialise the devices differently? Mine doesn't either, try standby? With this applied, I can resume from standby *without MSI* and the NIC still works. I haven't tested it without MSI and without the patch... mostly because I got a BUG when I tried to recompile. Attached. (The previous email has all the standby/resume log output.) Standby - resume from RAM is completely broken for my system. If I have MSI enabled, it still doesn't work.

Re: forcedeth 10de:0373 doesn't work on resume

Rafael J. Wysocki — 2008-08-18T21:13:09

[Must have missed this message.] I have a box with forcedeth that evidently works after a resume from hibernation. Unfortunately, so far I haven't been able to make the box resume from suspend to RAM. I'll do my best to try again tomorrow, but there's a little hope. :-( Simon, I'd prefer the full dmesg to the grepped forcedeth messages. I guess this was resume from suspend to RAM? It probably is. Thanks, Rafael

Re: forcedeth 10de:0373 doesn't work on resume

Andrew Morton — 2008-08-18T20:48:14

On Tue, 05 Aug 2008 20:29:44 +0100 Simon Arlott fire.lp0.eu> wrote: [two weeks pass...] That seems like a sensible change. Is it still broken in current kernels? Thanks.