Re: [PATCH 0/5] SATA/ACPI suspend/resume support

[Randy Dunlap <randy_d_dunlap@linux.intel.com>]

On 04 Jan 2006 10:31:23 +0900
Jae-hyeon Park <jhpark@tuhep.phys.tohoku.ac.jp> wrote:

> Randy Dunlap <randy_d_dunlap@linux.intel.com> writes:
> 
> > On 04 Jan 2006 07:40:00 +0900
> > Jae-hyeon Park <jhpark@tuhep.phys.tohoku.ac.jp> wrote:
> > > Randy Dunlap <randy_d_dunlap@linux.intel.com> writes:
> > > > On Tue, 3 Jan 2006 20:03:47 +0100
> > > > Jens Axboe <axboe@suse.de> wrote:
> > > > > On Tue, Jan 03 2006, Jens Axboe wrote:
> > > > > > On Tue, Jan 03 2006, Randy Dunlap wrote:
> > > > > > > On 30 Dec 2005 17:36:44 +0900
> > > > > > > Jae-hyeon Park <jhpark@tuhep.phys.tohoku.ac.jp> wrote:
> > > > > > > 
> > > > > > > > I tested kernel 2.6.15-rc7 with your patch applied on my ThinkPad X41.
> > > 
> > > My laptop is ThinkPad X41 Tablet, which has the same SATA controller
> > > as X41.
> > > 
> > > > > > > > While booting, it locks up after printing
> > > > > > > > 
> > > > > > > >     ata1: SATA max UDMA/133 cmd 0x1F0 ctl 0x3F6 bmdma 0x1810 irq 14
> > > > > > > >     ata1: dev 0 ATA-6, max UDMA/100, 117210240 sectors: LBA
> > > > > > > >     ata1(0): applying bridge limits
> > > > > > > >     ata1: dev 0 configured for UDMA/100
> > > > > > > > 
> > > > > > > > If I select CONFIG_SCSI_SATA_AHCI, then the kernel emits an oops and
> > > > > > > > panics during boot.
> > 
> > This certainly isn't the same failure that Jens saw.
> > 
> > Could there be any informative & helpful messages missing
> > before these messages, e.g., an assert() failure?
> 
> Let me copy as far as I can setting the video mode to 80x60.
> Scrolling back does not work after kernel panic.

Thanks.

> highmem bounce pool size: 64 pages
> Initializing Cryptographic API
> io scheduler noop registered
> io scheduler anticipatory registered
> serio: i8042 AUX port at 0x60,0x64 irq 12
> serio: i8042 KBD port at 0x60,0x64 irq 1
> acpi_bus-0201 [27] bus_set_power         : Device is not power manageable
> ahci: probe of 0000:00:1f.2 failed with error -12

-12 is -ENOMEM.  ahci can fail with -ENOMEM in 5 places.
3 of them are due to kmalloc() failures (real NOMEM).
1 is due to failure of dma_alloc_coherent() and
1 is due to failure of pci_iomap().

Have you seen any other shortage-of-memory conditions?


> ata1: SATA max UDMA/133 cmd 0x1F0 ctl 0x3F6 bmdma 0x1810 irq 14
> ata1: dev 0 ATA-6, max UDMA/100, 117210240 sectors: LBA
> ata1(0): applying bridge limits
> ata1: dev 0 configured for UDMA/100
> do_drive_set_taskfiles: unexpected GTF length (-1040297340)
> scsi0: ata_piix
>   Vendor: ATA       Model: HTC426060G9AT00   Rev: 00P3
>   Type:   Direct-Access                      ANSI SCSI revision: 05
> ata2: SATA max UDMA/133 cmd 0x170 ctl 0x376 bmdma 0x1818 irq 15
> ata2: disabling port
> scsi1 : ata_piix
> SCSI device sda: 117210240 512-byte hdwr sectors (60012 MB)
> SCSI device sda: drive cache: write back
> SCSI device sda: 117210240 512-byte hdwr sectors (60012 MB)
> SCSI device sda: drive cache: write back
>  sda:<1>Unable to handle kernel NULL pointer dereference at virtual address 00000172
>  ...
> 
> > I don't quite understand what CONFIG_SCSI_SATA_AHCI has to do
> > with this.  I see ata_interrupt() [from libata-core.c],
> > but ahci.c contains ahci_interrupt(), which isn't in this
> > stack dump.  Nevertheless, it looks like complete() is
> > being passed a NULL struct completion...  Maybe I'm misusing
> > the completion API and don't know it (?).
> > 
> > BTW, .config file, please.
> 
> .config was already included in my previous posting, with the name
> config-2.6.15-rc7.bz2.

Yes, thanks, I saw the acpidump but missed the config file.

> > > > New patch series is at
> > > >   http://www.xenotime.net/linux/SATA/2.6.15/
> > > > with a combined patch at
> > > >   http://www.xenotime.net/linux/SATA/2.6.15/libata-combine-2615.patch
> > > 
> > > With this patch series, resume still fails with snd_intel8x0m
> > > rmmod-ed.  Resume works if the snd_intel8x0m module has been inserted.
> > > I am curious about this, too.
> > 
> > Wow, interesting, but I have no idea about that.
> 
> As I said in a previous posting, 2.6.15-rc7 + git-libata-all.patch +
> libata_resume_fix.patch + libata_suspend-fix.patch +
> libata_suspend.patch, solved the problem, irrespective of the presence
> of snd_intel8x0m.  Among these four, libata_suspend.patch is Jens'
> patch.  So probably snd_intel8x0m issue is not related to your
> (modification of Jens') code.  Looks like one of the above four
> patches has the fix.
> 
> Another point that I observed is that your patch against 2.6.15-rc7,
> did not work together with suspend2 patch, even if snd_intel8x0m was
> inserted.  Again, the above combination works with suspend2.

---
~Randy