IDE Hard drive doesn't work after resume from S3 on Acer Aspire 1314

[Albert Vilella]

Suspend to RAM fails when I try to do a /etc/acpi/sleep.sh in my laptop.
Shuts X correctly, normally and really fast (1
or 2 seconds), all the hardware sleeping, the power indicator LED blinking.
But, when I press the "Power" key in my keyboard, or any of the special keys, X
session restores ok,
keyboard and mouse input ok, all fine but the hda seems to have trouble
recovering (hda led blinking hard), can't execute any command because the hda is
not accessible. Kernel outputs DMA timeouts for the hdd... After half a minute,
decides it can't restore, and goes for a controlled shutdown.

Steps to reproduce: Go "suspend" to S3. It goes to sleep mode, the power LED
blinks. If I press the power or specials button, it restores graphics, keyboard
and mouse
are working but hda has problems.

[Albert Vilella (avilella)]

changed the following options (in all combinations):

SAVE_VBE_STATE=true

VBESTATE=/var/lib/acpi-support/vbestate

POST_VIDEO=true

USE_DPMS=true

DOUBLE_CONSOLE_SWITCH=true

HIBERNATE_MODE=shutdown

DISABLE_DMA=true

STOP_SERVICES="mysql "

and same results.

[Albert Vilella (avilella)]

Created an attachment (id=1832)
dmidecode

[Albert Vilella (avilella)]

Created an attachment (id=1833)
lspci -vvv

[Albert Vilella (avilella)]

Created an attachment (id=1834)
dissasembled original acer aspire dsdt

[Albert Vilella (avilella)]

Won't find any problems with the reassembled dsdt table:

root@magneto:/home/avb # acpica-unix-20050309/compiler/iasl -tc dsdt.dsl

Intel ACPI Component Architecture
ASL Optimizing Compiler / AML Disassembler version 20050309 [Mar 24 2005]
Copyright (C) 2000 - 2005 Intel Corporation
Supports ACPI Specification Revision 3.0

ASL Input: dsdt.dsl - 2839 lines, 93278 bytes, 1282 keywords
AML Output: DSDT.aml - 11877 bytes 381 named objects 901 executable opcodes

Compilation complete. 0 Errors, 0 Warnings, 0 Remarks, 370 Optimizations
root@magneto:/home/avb #

[Albert Vilella (avilella)]

This is a laptop where the HD had to be changed for another about a year ago.
The original harddrive is not there then. I wonder if this has to do with any
miscommunication issue between the BIOS (dsdt) and the hardware/kernel...

Would a reassambled dsdt help? Note that the reassemblage does not give any
error neither warning.

[Albert Vilella (avilella)]

this happens both with DISABLE_DMA=true or commented

[Albert Vilella (avilella)]

Seems that some people in the kernel ml complained about this:

http://lkml.org/lkml/2003/6/10/320

same hardware here (see lscpi -vvv attachment)

[Albert Vilella (avilella)]

Correctly loading custom DSDT.aml:

DMI 2.3 present.
ACPI: RSDP (v000 OID_00 ) @ 0x000e5010
ACPI: RSDT (v001 INSYDE RSDT_000 0x00000001 _CSI 0x00010101) @ 0x1dfffbc0
ACPI: FADT (v001 INSYDE FACP_000 0x00000100 _CSI 0x00010101) @ 0x1dfffac0
ACPI: BOOT (v001 INSYDE SYS_BOOT 0x00000100 _CSI 0x00010101) @ 0x1dfffb50
ACPI: DBGP (v001 INSYDE DBGP_000 0x00000100 _CSI 0x00010101) @ 0x1dfffb80
ACPI: DSDT (v001 INSYDE KN266 0x00001000 MSFT 0x0100000e) @ 0x00000000
ACPI: PM-Timer IO Port: 0x1008
Built 1 zonelists
Kernel command line: root=/dev/hda1 ro quiet splash
Local APIC disabled by BIOS -- you can enable it with "lapic"
mapped APIC to ffffd000 (013c5000)
Initializing CPU#0
PID hash table entries: 2048 (order: 11, 32768 bytes)
Detected 1792.781 MHz processor.
Using pmtmr for high-res timesource
Console: colour VGA+ 80x25
Dentry cache hash table entries: 65536 (order: 6, 262144 bytes)
Inode-cache hash table entries: 32768 (order: 5, 131072 bytes)
Memory: 479296k/491456k available (1594k kernel code, 11564k reserved, 721k
data, 164k init, 0k highmem)
Checking if this processor honours the WP bit even in supervisor mode... Ok.
Calibrating delay loop... 3538.94 BogoMIPS (lpj=1769472)
Security Framework v1.0.0 initialized
SELinux: Disabled at boot.
Mount-cache hash table entries: 512 (order: 0, 4096 bytes)
CPU: After generic identify, caps: 0383f9ff c1cbf9ff 00000000 00000000 00000000
00000000
CPU: After vendor identify, caps: 0383f9ff c1cbf9ff 00000000 00000000 00000000
00000000
CPU: L1 I Cache: 64K (64 bytes/line), D cache 64K (64 bytes/line)
CPU: L2 Cache: 512K (64 bytes/line)
CPU: After all inits, caps: 0383f9ff c1cbf9ff 00000000 00000020 00000000 00000000
Intel machine check architecture supported.
Intel machine check reporting enabled on CPU#0.
CPU: AMD Athlon(tm) XP 2400+ stepping 00
Enabling fast FPU save and restore... done.
Enabling unmasked SIMD FPU exception support... done.
Checking 'hlt' instruction... OK.
ACPI: Looking for DSDT in initrd... found (at offset 4648980)!
 found customized DSDT with 11857 bytes!
ACPI: Using customized DSDT
    ACPI-0294: *** Info: Table [DSDT] replaced by host OS
ACPI: setting ELCR to 0200 (from 0e20)
checking if image is initramfs...it isn't (bad gzip magic numbers); looks like
an initrd

[Albert Vilella (avilella)]

Created an attachment (id=1836)
ubuntu hardware db data

[Albert Vilella (avilella)]

I have also tried several combinations of "apm=off" "ide=nodma" "noapic" "nolapic"
and they also result in the hda problem _or_ in recovering to the startup bios
point,
where a normal bootup, instead of a wakeup, is triggered... weird.

[Matthew Garrett (mjg59)]

Linux currently has no real support for setting up IDE interfaces on resume.
Some machines are kind enough to set the IDE interface up themselves, but on
others we're doomed to failure. I'm looking into implementing this, but it won't
happen until some time after Hoary.

[Albert Vilella (avilella)]

laptop mode can stop and restart the HD spinning while on batteries with no
problems - not sure if that adds any useful info

[Albert Vilella (avilella)]

Matthew, is anything being done kernel-wise or ubuntu-wise around this issue?
Can I provide any other info?

[Matthew Garrett (mjg59)]

I haven't managed to get the upstream kernel guys interested in this, which
probably means I'll need to look at writing it myself at some stage. I'm kind of
busy at the moment, so don't expect anything too quickly...

[Albert Vilella (avilella)]

My bug report in redhat bug tracker was already sent upstream by Alan Cox:
https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=155102

So I suppose that means it already is in the "to do" list in the upstream bug
tracker... (kernel?)

[Matthew Garrett (mjg59)]

The 2.6.12 kernel in Breezy should fix this issue (though please wait until the
version that should be out on Monday - there are currently other power
management issues with the kernel in Breezy)

[Albert Vilella (avilella)]

hibernate (suspend-to-disk) works fine under Breezy Col3

sleep (suspend-to-ram, i.e. S3) same problems with 2.6.12-7-k7 kernel as of Monday.
I assume this is not yet the kernel version that should work, right?

[Matthew Garrett (mjg59)]

Hmm. No, that one includes the code that I'd hoped would fix that issue, so it
sounds like there may be something else going on. I'll look into this further.

[Dr. Stefan Schimanski (sttts)]

Same problem here with Acer TM800. Any progress on this issue?

[Albert Vilella (avilella)]

It wooooorks!!!

I installed official Breezy today and it works nicely!!!

Thank you Matthew and company for the great job!!!

[Dennis Kaarsemaker (dennis)]

Good to hear :)

[Albert Vilella (avilella)]

Well, it seems that it plays well _sometimes_:

If the time lapse between the sleep and the wake-up is prolonged, it won't wake
up properly: it simply reinitiates from the BIOS as if it where turned on or
rebooted.

Could this have any relation with laptop_mode?

Is there anything I can test to bring more information?