Bugzilla – Bug 145591
FATE #151275: suspend to RAM (S3) does not work on many HP notebooks
Last modified: 2006-07-14 15:13:48 UTC
newer HP notebooks do not recover from suspend to RAM because of missing ACPI handling in the IDE layer. There is a patch around which fixes this: http://thread.gmane.org/gmane.linux.ide/6802 See also bug #142375 for a similar bug with SATA.
Created attachment 65065 [details] the patch, adopted to 2.6.16rc1 i had to add 3 additional EXPORT_SYMBOLs and am not sure that they are in the best possible place.
Is that patch (or its equivalent) in mainline? First it looks slightly ugly (#includes should not be protected by #ifdefs, clean the code not to need the protection), then I'm afraid it will break some machines. What machines are we trying to fix? I seen problems in compaq evo 520c and nx5000 (IIRC); and those were way before reaching ide resume, and probably in BIOS code.
Scary patch indeed. Jens, any opinion on this one?
(In reply to comment #2) > Is that patch (or its equivalent) in mainline? No. It was discussed around 2005-12-06 on lkml and linux-ide, the patch was brought up again by Shaohua Li. There is also some discussion on http://bugzilla.kernel.org/show_bug.cgi?id=2039 and http://bugzilla.kernel.org/show_bug.cgi?id=5604 IIRC some consensus was that "#ifdef CONFIG_ACPI" code did not belong into ide core and finally hch shot it down (or was this the first round of the SATA code? not sure). > First it looks slightly ugly (#includes should not be protected by #ifdefs, This was also mentioned in the thread on linux-ide. Also, the +#include <acpi/acpixf.h> is probably bogus - i added that before i realized that i need to do the EXPORT_SYMBOL stuff in the ACPI code. > clean the code not to need the protection), then I'm afraid it will break some > machines. Ubuntu ships this for a long time, reportedly without breakage. > What machines are we trying to fix? I seen problems in compaq evo 520c and > nx5000 (IIRC); and those were way before reaching ide resume, and probably in > BIOS code. nx5000 and friends were fixed by "the great ACPI interpreter fixup" (around ~2.6.9 IIRC) which got many machien to resume suddenly, i believe i have seen nx5000 work almost ootb in 9.2 or 9.3. The ones failing now are the newer machines - n[cx]61xx, n[cx]62xx, n[cx]82xx. Those resume fine (video is hosed), but you can no longer access the disk - all you get is a timeout - almost feels like the early sata suspend stuff. With this patch, "vga=0" and "vbetool vbestate restore" after resume, my nc6230 works just fine. I am adding Shaohua Li to Cc:, maybe he can shed some light on this issue.
My main problem with the patch is that I don't understand what it does. It doesn't actually submit any commands to the drive, so honestly right now I don't know why it makes a difference. Perhaps Intel can enlighten me?
Adding those PMs that could benefit from this patch ;-)
ACPI spec requires IDE driver executes some ACPI methods to save/restore IDE status. The patch is just a workaround. Per the discussion in ide maillist, there are many things to do to improve the patch.
Created attachment 71328 [details] ide-acpi-support.patch Patch to add ACPI support for IDE drives.
The above patch is a modification of Randy Dunlaps patch for libata. We should consider including that one instead of the one from comment #1 as we (or Randy) could integrate both so that we only have to maintain one codebase.
*** Bug 155304 has been marked as a duplicate of this bug. ***
Could you run the patch through linux-kernel for comments? At least = 0 initializers should be stripped, ide-acpi.c needs GPL, you are using DEBPRINT for routine debugging (probably should be removed prior to merge) and error conditions, ...
Randy, are you still working on this kind of issue now that you are no longer with Intel (it's fine if not, just need to determine who will take over the submission of this patch.)
Yes, I hope to post a libata patch very soon (not for drivers/ide/, like Hannes posted recently).
Created attachment 72776 [details] ide-acpi-support Updated version of the IDE ACPI support. Includes suggestions made by Randy on linux-ide as well as some module parameters to tweak operations.
And with the above patch even seife's HP Laptop resumes properly. So this would warrant to set this bug to 'FIXED', especially as I've added the patch to our kernel CVS.
Was patch submitted upstream?
Yes, see thread '[PATCH] ACPI support for IDE drives' on linux-ide. Not that I got any response, but hey, you can't have everything. Pavel, any comments on the patch are greatly appreciated.
Will the patch be on beta8? I have an HP that doesn't wake up, too, and I'd like to verify ASAP.
yes, it is.
Closed.