Debian Bug report logs - #518750
pommed: cannot mmap memory for GMA950/GMA965

Reply or subscribe to this bug.

Toggle useless messages

Message #5 received at submit@bugs.debian.org (full text, mbox, reply):

From: Filippo Giunchedi <filippo@debian.org>

To: Debian Bug Tracking System <submit@bugs.debian.org>

Subject: pommed: cannot mmap memory for GMA950/GMA965

Date: Sun, 08 Mar 2009 12:34:07 +0100

[Message part 1 (text/plain, inline)]

Package: pommed
Version: 1.25~dfsg-1
Severity: normal

[I'm not entirely sure this is related to pommed, reassing as needed]

Hi,
pommed doesn't start while failing to mmap the video memory:

DMI vendor name: [Apple Inc.]
DMI product name: [MacBook3,1]
I: DMI machine check: running on a MacBook3,1
System: Linux 2.6.28.6 x86_64E: mmap failed: Invalid argument
E: Could not map GMA950/GMA965 memory
E: LCD backlight probe failed, check debug output

attached there's the -d output plus the strace

thanks,
filippo

-- System Information:
Debian Release: squeeze/sid
  APT prefers unstable
  APT policy: (990, 'unstable')
Architecture: amd64 (x86_64)

Kernel: Linux 2.6.28.6 (SMP w/2 CPU cores)
Locale: LANG=en_US.UTF-8, LC_CTYPE=en_US.UTF-8 (charmap=UTF-8)
Shell: /bin/sh linked to /bin/bash

Versions of packages pommed depends on:
ii  eject           2.1.5+deb1+cvs20081104-5 ejects CDs and operates CD-Changer
ii  libasound2      1.0.19-1                 shared library for ALSA applicatio
ii  libaudiofile0   0.2.6-7                  Open-source version of SGI's audio
ii  libc6           2.9-4                    GNU C Library: Shared libraries
ii  libconfuse0     2.6-2                    Library for parsing configuration 
ii  libdbus-1-3     1.2.12-1                 simple interprocess messaging syst
ii  libpci3         1:3.1.2-1                Linux PCI Utilities (shared librar

Versions of packages pommed recommends:
ii  dbus                          1.2.12-1   simple interprocess messaging syst

pommed suggests no packages.

-- no debconf information

[pommed_strace.log (text/x-c, attachment)]

[pommed_debug.log (text/plain, attachment)]

Message #10 received at 518750@bugs.debian.org (full text, mbox, reply):

From: Julien BLACHE <jblache@debian.org>

To: Filippo Giunchedi <filippo@debian.org>

Cc: 518750@bugs.debian.org

Subject: Re: Bug#518750: pommed: cannot mmap memory for GMA950/GMA965

Date: Sun, 08 Mar 2009 18:47:07 +0100

Filippo Giunchedi <filippo@debian.org> wrote:

Hi,

> I: DMI machine check: running on a MacBook3,1
> System: Linux 2.6.28.6 x86_64E: mmap failed: Invalid argument
> E: Could not map GMA950/GMA965 memory
> E: LCD backlight probe failed, check debug output

Hmm, strange. lspci -v for your graphic card, please?

Any kernel messages in addition to this?

JB.

-- 
 Julien BLACHE - Debian & GNU/Linux Developer - <jblache@debian.org> 
 
 Public key available on <http://www.jblache.org> - KeyID: F5D6 5169 
 GPG Fingerprint : 935A 79F1 C8B3 3521 FD62 7CC7 CD61 4FD7 F5D6 5169 

Message #15 received at 518750@bugs.debian.org (full text, mbox, reply):

From: Filippo Giunchedi <filippo@debian.org>

To: Julien BLACHE <jblache@debian.org>

Cc: 518750@bugs.debian.org

Subject: Re: Bug#518750: pommed: cannot mmap memory for GMA950/GMA965

Date: Sun, 8 Mar 2009 20:23:28 +0100

[Message part 1 (text/plain, inline)]

On Sun, Mar 08, 2009 at 06:47:07PM +0100, Julien BLACHE wrote:
> Filippo Giunchedi <filippo@debian.org> wrote:
> 
> Hi,
> 
> > I: DMI machine check: running on a MacBook3,1
> > System: Linux 2.6.28.6 x86_64E: mmap failed: Invalid argument
> > E: Could not map GMA950/GMA965 memory
> > E: LCD backlight probe failed, check debug output
> 
> Hmm, strange. lspci -v for your graphic card, please?

strange indeed, lspci attached

> 
> Any kernel messages in addition to this?

no kernel messages, I'm rather puzzled as apparently nothing has changed since
last reboot (I did a software update in OSX but I doubt this is relevant)

thanks,
filippo
--
Filippo Giunchedi - http://esaurito.net - 0x6B79D401

If there is any better use for being famous and respected than using
that status to question orthodoxy, I haven't found it yet.
-- Eric S. Raymond

[lspci (text/plain, attachment)]

Message #20 received at 518750@bugs.debian.org (full text, mbox, reply):

From: Julien BLACHE <jblache@debian.org>

To: Filippo Giunchedi <filippo@debian.org>

Cc: 518750@bugs.debian.org

Subject: Re: Bug#518750: pommed: cannot mmap memory for GMA950/GMA965

Date: Sun, 08 Mar 2009 20:57:14 +0100

Filippo Giunchedi <filippo@debian.org> wrote:

Hi,

>> Any kernel messages in addition to this?
>
> no kernel messages, I'm rather puzzled as apparently nothing has changed since
> last reboot (I did a software update in OSX but I doubt this is relevant)

Did you update your Linux kernel? Is your kernel built with strict
access control for /dev/mem?

JB.

-- 
 Julien BLACHE - Debian & GNU/Linux Developer - <jblache@debian.org> 
 
 Public key available on <http://www.jblache.org> - KeyID: F5D6 5169 
 GPG Fingerprint : 935A 79F1 C8B3 3521 FD62 7CC7 CD61 4FD7 F5D6 5169 

Message #25 received at 518750@bugs.debian.org (full text, mbox, reply):

From: Filippo Giunchedi <filippo@debian.org>

To: Julien BLACHE <jblache@debian.org>

Cc: 518750@bugs.debian.org

Subject: Re: Bug#518750: pommed: cannot mmap memory for GMA950/GMA965

Date: Sun, 8 Mar 2009 22:37:07 +0100

On Sun, Mar 08, 2009 at 08:57:14PM +0100, Julien BLACHE wrote:
> Filippo Giunchedi <filippo@debian.org> wrote:
> 
> Hi,
> 
> >> Any kernel messages in addition to this?
> >
> > no kernel messages, I'm rather puzzled as apparently nothing has changed since
> > last reboot (I did a software update in OSX but I doubt this is relevant)
> 
> Did you update your Linux kernel? Is your kernel built with strict
> access control for /dev/mem?

CONFIG_STRICT_DEVMEM is disabled, I'll retry with a different kernel and the
debian one though.

thanks,
filippo
--
Filippo Giunchedi - http://esaurito.net - 0x6B79D401

God may not play dice with the universe, but something strange is going on with
the prime numbers.
-- Paul Erdos

Message #30 received at 518750@bugs.debian.org (full text, mbox, reply):

From: Filippo Giunchedi <filippo@debian.org>

To: Julien BLACHE <jblache@debian.org>

Cc: 518750@bugs.debian.org

Subject: Re: Bug#518750: pommed: cannot mmap memory for GMA950/GMA965

Date: Mon, 9 Mar 2009 08:07:41 +0100

[Message part 1 (text/plain, inline)]

On Sun, Mar 08, 2009 at 08:57:14PM +0100, Julien BLACHE wrote:
> Filippo Giunchedi <filippo@debian.org> wrote:
> 
> Hi,
> 
> >> Any kernel messages in addition to this?
> >
> > no kernel messages, I'm rather puzzled as apparently nothing has changed since
> > last reboot (I did a software update in OSX but I doubt this is relevant)
> 
> Did you update your Linux kernel? Is your kernel built with strict
> access control for /dev/mem?

Changing the kernel doesn't affect the result, straces attached.
Also FWIW gnome-power-manager is able to correctly increase the brightness while
launched.

filippo
--
Filippo Giunchedi - http://esaurito.net - 0x6B79D401

If you could say it in words there would be no reason to paint.
-- Edward Hopper

[pommed_2.6.28-1-amd64.strace (text/plain, attachment)]

[pommed_2.6.28.7.strace (text/plain, attachment)]

Message #35 received at 518750@bugs.debian.org (full text, mbox, reply):

From: Julien BLACHE <jblache@debian.org>

To: Filippo Giunchedi <filippo@debian.org>

Cc: 518750@bugs.debian.org

Subject: Re: Bug#518750: pommed: cannot mmap memory for GMA950/GMA965

Date: Mon, 09 Mar 2009 12:16:05 +0100

Filippo Giunchedi <filippo@debian.org> wrote:

Hi,

> Changing the kernel doesn't affect the result, straces attached.

I'd like to get the complete pommed output, including debug on stderr,
please :) That's actually more useful than the strace.

> Also FWIW gnome-power-manager is able to correctly increase the
> brightness while launched.

That probably goes through xrandr and X. What do you have in
/sys/class/backlight? Maybe a backlight driver made it in the kernel
which I missed.

JB.

-- 
 Julien BLACHE - Debian & GNU/Linux Developer - <jblache@debian.org> 
 
 Public key available on <http://www.jblache.org> - KeyID: F5D6 5169 
 GPG Fingerprint : 935A 79F1 C8B3 3521 FD62 7CC7 CD61 4FD7 F5D6 5169 

Message #40 received at 518750@bugs.debian.org (full text, mbox, reply):

From: Filippo Giunchedi <filippo@debian.org>

To: Julien BLACHE <jblache@debian.org>

Cc: 518750@bugs.debian.org

Subject: Re: Bug#518750: pommed: cannot mmap memory for GMA950/GMA965

Date: Mon, 9 Mar 2009 17:41:10 +0100

[Message part 1 (text/plain, inline)]

On Mon, Mar 09, 2009 at 12:16:05PM +0100, Julien BLACHE wrote:
> Filippo Giunchedi <filippo@debian.org> wrote:
> 
> Hi,
> 
> > Changing the kernel doesn't affect the result, straces attached.
> 
> I'd like to get the complete pommed output, including debug on stderr,
> please :) That's actually more useful than the strace.

okay, attached for 2.6.28.7

> 
> > Also FWIW gnome-power-manager is able to correctly increase the
> > brightness while launched.
> 
> That probably goes through xrandr and X. What do you have in
> /sys/class/backlight? Maybe a backlight driver made it in the kernel
> which I missed.

$ ls /sys/class/backlight/
$ 

this is with 2.6.28.7, config attached just in case.

also,

$ dpkg -l '*xorg*' | grep ^ii
ii  xserver-xorg                                             1:7.4~5 the X.Org X server
ii  xserver-xorg-core                                        2:1.5.99.902-1 Xorg X server - core server
ii  xserver-xorg-input-evdev                                 1:2.1.1-1 X.Org X server -- evdev input driver
ii  xserver-xorg-input-kbd                                   1:1.3.2-2 X.Org X server -- keyboard input driver
ii  xserver-xorg-input-mouse                                 1:1.4.0-1 X.Org X server -- mouse input driver
ii  xserver-xorg-input-synaptics                             0.99.3-3 Synaptics TouchPad driver for X.Org server
ii  xserver-xorg-video-intel                                 2:2.6.1-1 X.Org X server -- Intel i8xx, i9xx display d
ii  xserver-xorg-video-vesa                                  1:2.1.0-1 X.Org X server -- VESA display driver

$ dpkg -l '*randr*' | grep ^ii
ii  libxcb-randr0                                            1.1.93-1 X C Binding, randr extension
ii  libxrandr-dev                                            2:1.2.99.4-2 X11 RandR extension library (development hea
ii  libxrandr2                                               2:1.2.99.4-2 X11 RandR extension library
ii  x11proto-randr-dev                                       1.2.99.3-2 X11 RandR extension wire protocol

thanks,
filippo
--
Filippo Giunchedi - http://esaurito.net - 0x6B79D401

We are human because our ancestors learned to share their food and their skills
in an honored network of obligation.
-- Richard Leakey 

[pommed_2.6.28.7.log (text/plain, attachment)]

[config-2.6.28.7 (text/plain, attachment)]

Message #45 received at 518750@bugs.debian.org (full text, mbox, reply):

From: Julien BLACHE <jblache@debian.org>

To: Filippo Giunchedi <filippo@debian.org>

Cc: 518750@bugs.debian.org

Subject: Re: Bug#518750: pommed: cannot mmap memory for GMA950/GMA965

Date: Mon, 09 Mar 2009 18:16:21 +0100

Filippo Giunchedi <filippo@debian.org> wrote:

Hi,

>> I'd like to get the complete pommed output, including debug on stderr,
>> please :) That's actually more useful than the strace.
>
> okay, attached for 2.6.28.7

Thanks. The mmap fails for some reason that's not immediately obvious
:/ Did you ever have a working pommed?

>> That probably goes through xrandr and X. What do you have in
>> /sys/class/backlight? Maybe a backlight driver made it in the kernel
>> which I missed.
>
> $ ls /sys/class/backlight/
> $ 

OK that's a no...

The problem looks specific to the GMA support, as I'm running a 2.6.28
x86_64 kernel here on my MBP and pommed works fine with my Radeon card
:|

JB.

-- 
 Julien BLACHE - Debian & GNU/Linux Developer - <jblache@debian.org> 
 
 Public key available on <http://www.jblache.org> - KeyID: F5D6 5169 
 GPG Fingerprint : 935A 79F1 C8B3 3521 FD62 7CC7 CD61 4FD7 F5D6 5169 

Message #50 received at 518750@bugs.debian.org (full text, mbox, reply):

From: Filippo Giunchedi <filippo@debian.org>

To: Julien BLACHE <jblache@debian.org>

Cc: 518750@bugs.debian.org

Subject: Re: Bug#518750: pommed: cannot mmap memory for GMA950/GMA965

Date: Mon, 9 Mar 2009 18:22:30 +0100

On Mon, Mar 09, 2009 at 06:16:21PM +0100, Julien BLACHE wrote:
> Filippo Giunchedi <filippo@debian.org> wrote:
> 
> Hi,
> 
> >> I'd like to get the complete pommed output, including debug on stderr,
> >> please :) That's actually more useful than the strace.
> >
> > okay, attached for 2.6.28.7
> 
> Thanks. The mmap fails for some reason that's not immediately obvious
> :/ Did you ever have a working pommed?

sure, it's been working flawlessly since ever! (implicit "thanks for pommed" here)

That's why I'm rather puzzled at this... I'm not sure what has changed or where
to look. I'm starting to think that I might be missing something obvious.

filippo
--
Filippo Giunchedi - http://esaurito.net - 0x6B79D401

Beauty isn't worth thinking about; what's important is your mind.
You don't want a fifty-dollar haircut on a fifty-cent head.
-- Garrison Keillor

Message #55 received at 518750@bugs.debian.org (full text, mbox, reply):

From: Julien BLACHE <jblache@debian.org>

To: Filippo Giunchedi <filippo@debian.org>

Cc: 518750@bugs.debian.org

Subject: Re: Bug#518750: pommed: cannot mmap memory for GMA950/GMA965

Date: Mon, 09 Mar 2009 21:04:28 +0100

Filippo Giunchedi <filippo@debian.org> wrote:

Hi,

>> Thanks. The mmap fails for some reason that's not immediately obvious
>> :/ Did you ever have a working pommed?
>
> sure, it's been working flawlessly since ever! (implicit "thanks for pommed" here)

OK, that's a good data point (and you're welcome ;))

> That's why I'm rather puzzled at this... I'm not sure what has changed or where
> to look. I'm starting to think that I might be missing something obvious.

That was your lspci output:

00:02.0 VGA compatible controller: Intel Corporation Mobile GM965/GL960 Integrated Graphics Controller (rev 03) (prog-if 00 [VGA controller])
	Subsystem: Apple Computer Inc. Device 00a1
	Flags: bus master, fast devsel, latency 0, IRQ 315
	Memory at d0100000 (64-bit, non-prefetchable) [size=1M]
	Memory at c0000000 (64-bit, prefetchable) [size=256M]
	I/O ports at 3110 [size=8]
	Expansion ROM at <unassigned> [disabled]
	Capabilities: [90] MSI: Mask- 64bit- Count=1/1 Enable+
	Capabilities: [d0] Power Management version 3
	Kernel modules: intelfb

00:02.1 Display controller: Intel Corporation Mobile GM965/GL960 Integrated Graphics Controller (rev 03)
	Subsystem: Apple Computer Inc. Device 00a1
	Flags: bus master, fast devsel, latency 0
	Memory at d0200000 (64-bit, non-prefetchable) [size=1M]
	Capabilities: [d0] Power Management version 3

Compare with:

<http://launchpadlibrarian.net/16481998/lspci-vvnn.log>

The memory regions are the same except for one bit. Something changed
at some level and the memory regions aren't reported in the same way
anymore. (wonder if it's linked to the OS X update?)

Can you send me your dmesg? I'd like to see what the PCI layer has to
tell about that. (mainly to see if using resource files under /sys
would work as opposed to using /dev/mem)

JB.

-- 
 Julien BLACHE - Debian & GNU/Linux Developer - <jblache@debian.org> 
 
 Public key available on <http://www.jblache.org> - KeyID: F5D6 5169 
 GPG Fingerprint : 935A 79F1 C8B3 3521 FD62 7CC7 CD61 4FD7 F5D6 5169 

Message #60 received at 518750@bugs.debian.org (full text, mbox, reply):

From: Julien BLACHE <jblache@debian.org>

To: Filippo Giunchedi <filippo@debian.org>

Cc: 518750@bugs.debian.org

Subject: Re: Bug#518750: pommed: cannot mmap memory for GMA950/GMA965

Date: Mon, 09 Mar 2009 21:09:41 +0100

Filippo Giunchedi <filippo@debian.org> wrote:

Hi,

> That's why I'm rather puzzled at this... I'm not sure what has changed or where
> to look. I'm starting to think that I might be missing something obvious.

Oh, and, if you could try to downgrade libpci3 and pciutils to the
Lenny version, that could help, too. Compare the output of lspci from
both versions. If they differ, we have a smoking gun.

Given that a bug in pciutils is a lot more likely than anything due to
your OS X update...

JB.

-- 
 Julien BLACHE - Debian & GNU/Linux Developer - <jblache@debian.org> 
 
 Public key available on <http://www.jblache.org> - KeyID: F5D6 5169 
 GPG Fingerprint : 935A 79F1 C8B3 3521 FD62 7CC7 CD61 4FD7 F5D6 5169 

Message #65 received at 518750@bugs.debian.org (full text, mbox, reply):

From: Filippo Giunchedi <filippo@debian.org>

To: Julien BLACHE <jblache@debian.org>

Cc: 518750@bugs.debian.org

Subject: Re: Bug#518750: pommed: cannot mmap memory for GMA950/GMA965

Date: Mon, 9 Mar 2009 21:38:39 +0100

[Message part 1 (text/plain, inline)]

On Mon, Mar 09, 2009 at 09:04:28PM +0100, Julien BLACHE wrote:
> Compare with:
> 
> <http://launchpadlibrarian.net/16481998/lspci-vvnn.log>
> 
> The memory regions are the same except for one bit. Something changed
> at some level and the memory regions aren't reported in the same way
> anymore. (wonder if it's linked to the OS X update?)

intriguing, FWIW these are the latest two updates:

2009-02-15 12:15:59 +0100: Installed "Java for Mac OS X 10.5 Update 3" (1.0)
2009-02-15 12:16:06 +0100: Installed "Pages Update" (3.0.3)
2009-02-15 12:16:17 +0100: Installed "Keynote Update" (4.0.4)
2009-02-15 12:16:32 +0100: Installed "Security Update 2009-001" (1.0)
2009-02-15 12:17:00 +0100: Installed "Numbers Update" (1.0.3)
2009-03-08 11:24:44 +0100: Installed "AirPort Utility Software Update 2009-001" (5.4.1)
2009-03-08 11:24:48 +0100: Installed "Digital Camera Raw Compatibility Update" (2.5)
2009-03-08 11:25:00 +0100: Installed "iLife Support" (9.0.1)
2009-03-08 11:25:02 +0100: Installed "AirPort Client Update 2009-001" (1.0)
2009-03-08 11:25:18 +0100: Installed "iPhoto Update" (8.0.1)

> 
> Can you send me your dmesg? I'd like to see what the PCI layer has to
> tell about that. (mainly to see if using resource files under /sys
> would work as opposed to using /dev/mem)

sure, attached.

thanks,
filippo
--
Filippo Giunchedi - http://esaurito.net - 0x6B79D401

I find television very educating. Every time somebody turns on the
set, I go into the other room and read a book.
-- Groucho Marx

[kern.log (text/plain, attachment)]

Message #70 received at 518750@bugs.debian.org (full text, mbox, reply):

From: Julien BLACHE <jblache@debian.org>

To: Filippo Giunchedi <filippo@debian.org>

Cc: 518750@bugs.debian.org

Subject: Re: Bug#518750: pommed: cannot mmap memory for GMA950/GMA965

Date: Mon, 09 Mar 2009 21:53:52 +0100

Filippo Giunchedi <filippo@debian.org> wrote:

Hi,

>> Can you send me your dmesg? I'd like to see what the PCI layer has to
>> tell about that. (mainly to see if using resource files under /sys
>> would work as opposed to using /dev/mem)
>
> sure, attached.

Hmm, the kernel agrees with libpci.

Can you copy the content of /sys/bus/pci/devices/0000:00:02.0/resource ?

JB.

-- 
 Julien BLACHE - Debian & GNU/Linux Developer - <jblache@debian.org> 
 
 Public key available on <http://www.jblache.org> - KeyID: F5D6 5169 
 GPG Fingerprint : 935A 79F1 C8B3 3521 FD62 7CC7 CD61 4FD7 F5D6 5169 

Message #75 received at 518750@bugs.debian.org (full text, mbox, reply):

From: Filippo Giunchedi <filippo@debian.org>

To: Julien BLACHE <jblache@debian.org>

Cc: 518750@bugs.debian.org

Subject: Re: Bug#518750: pommed: cannot mmap memory for GMA950/GMA965

Date: Mon, 9 Mar 2009 22:02:02 +0100

On Mon, Mar 09, 2009 at 09:53:52PM +0100, Julien BLACHE wrote:
> Filippo Giunchedi <filippo@debian.org> wrote:
> 
> Hi,
> 
> >> Can you send me your dmesg? I'd like to see what the PCI layer has to
> >> tell about that. (mainly to see if using resource files under /sys
> >> would work as opposed to using /dev/mem)
> >
> > sure, attached.
> 
> Hmm, the kernel agrees with libpci.
> 
> Can you copy the content of /sys/bus/pci/devices/0000:00:02.0/resource ?

$ cat '/sys/bus/pci/devices/0000:00:02.0/resource'
0x00000000d0100000 0x00000000d01fffff 0x0000000000020204
0x0000000000000000 0x0000000000000000 0x0000000000000000
0x00000000c0000000 0x00000000cfffffff 0x000000000002120c
0x0000000000000000 0x0000000000000000 0x0000000000000000
0x0000000000003110 0x0000000000003117 0x0000000000020101
0x0000000000000000 0x0000000000000000 0x0000000000000000
0x0000000000000000 0x0000000000000000 0x0000000000000002

filippo
--
Filippo Giunchedi - http://esaurito.net - 0x6B79D401

I never forget a face, but in your case I'll be glad to make an exception.
-- Groucho Marx

Message #80 received at 518750@bugs.debian.org (full text, mbox, reply):

From: Filippo Giunchedi <filippo@debian.org>

To: Julien BLACHE <jblache@debian.org>

Cc: 518750@bugs.debian.org

Subject: Re: Bug#518750: pommed: cannot mmap memory for GMA950/GMA965

Date: Mon, 9 Mar 2009 23:01:54 +0100

[Message part 1 (text/plain, inline)]

On Mon, Mar 09, 2009 at 09:38:39PM +0100, Filippo Giunchedi wrote:
> > 
> > Can you send me your dmesg? I'd like to see what the PCI layer has to
> > tell about that. (mainly to see if using resource files under /sys
> > would work as opposed to using /dev/mem)
> 
> sure, attached.

attached there's a boot log from a working kernel some time ago.

filippo
--
Filippo Giunchedi - http://esaurito.net - 0x6B79D401

We are human because our ancestors learned to share their food and their skills
in an honored network of obligation.
-- Richard Leakey 

[kern.log.2 (text/plain, attachment)]

Message #85 received at 518750@bugs.debian.org (full text, mbox, reply):

From: Julien BLACHE <jblache@debian.org>

To: Filippo Giunchedi <filippo@debian.org>

Cc: 518750@bugs.debian.org

Subject: Re: Bug#518750: pommed: cannot mmap memory for GMA950/GMA965

Date: Mon, 09 Mar 2009 23:08:03 +0100

Filippo Giunchedi <filippo@debian.org> wrote:

Hi,

> attached there's a boot log from a working kernel some time ago.

Hmm. The PCI mapping is the same, so that's now ruled out.

JB.

-- 
 Julien BLACHE - Debian & GNU/Linux Developer - <jblache@debian.org> 
 
 Public key available on <http://www.jblache.org> - KeyID: F5D6 5169 
 GPG Fingerprint : 935A 79F1 C8B3 3521 FD62 7CC7 CD61 4FD7 F5D6 5169 

Message #90 received at 518750@bugs.debian.org (full text, mbox, reply):

From: Julien BLACHE <jblache@debian.org>

To: Filippo Giunchedi <filippo@debian.org>

Cc: 518750@bugs.debian.org

Subject: Re: Bug#518750: pommed: cannot mmap memory for GMA950/GMA965

Date: Mon, 09 Mar 2009 23:21:20 +0100

Filippo Giunchedi <filippo@debian.org> wrote:

Hi,

> Changing the kernel doesn't affect the result, straces attached.

From the strace:

mmap(NULL, 1048576, PROT_READ|PROT_WRITE, MAP_SHARED, 4, 0xd0100004) = -1 EINVAL (Invalid argument)

The address should be 0xd010000*0* here, as we see in the lspci
output. That value is taken verbatim by pommed from libpci, so libpci
is giving us a bogus value.

Filippo confirms that pommed works fine with the libpci3 from Lenny.

JB.

-- 
 Julien BLACHE - Debian & GNU/Linux Developer - <jblache@debian.org> 
 
 Public key available on <http://www.jblache.org> - KeyID: F5D6 5169 
 GPG Fingerprint : 935A 79F1 C8B3 3521 FD62 7CC7 CD61 4FD7 F5D6 5169 

Message #95 received at 518750@bugs.debian.org (full text, mbox, reply):

From: Julien BLACHE <jblache@debian.org>

To: anibal@debian.org

Cc: 518750@bugs.debian.org, control@bugs.debian.org

Subject: Re: Bug#518750: pommed: cannot mmap memory for GMA950/GMA965

Date: Mon, 09 Mar 2009 23:50:27 +0100

reassign 518750 libpci3 1:3.1.2-1
thanks

Filippo Giunchedi <filippo@debian.org> wrote:

Hi Anibal,

See the bug log for the details, summary below.

> mmap(NULL, 1048576, PROT_READ|PROT_WRITE, MAP_SHARED, 4, 0xd0100004) = -1 EINVAL (Invalid argument)

This is a mmap call issued by pommed, with the address taken verbatim
from libpci.

As you can see in the lspci output, the address is actually
0xd010000*0*; the 4 we see here is probably the flag indicating a
64bit BAR that is not cleared by libpci.

libpci3 from Lenny does pass 0xd0100000 as expected. It looks like a
subtle bug in libpci; can you investigate this issue and confirm?

Thanks,

JB.

-- 
 Julien BLACHE - Debian & GNU/Linux Developer - <jblache@debian.org> 
 
 Public key available on <http://www.jblache.org> - KeyID: F5D6 5169 
 GPG Fingerprint : 935A 79F1 C8B3 3521 FD62 7CC7 CD61 4FD7 F5D6 5169 

Message #102 received at 518750@bugs.debian.org (full text, mbox, reply):

From: Aníbal Monsalve Salazar <anibal@debian.org>

To: Martin Mares <mj@ucw.cz>

Cc: Julien BLACHE <jblache@debian.org>, Filippo Giunchedi <filippo@debian.org>, 518750@bugs.debian.org

Subject: Re: Bug#518750: pommed: cannot mmap memory for GMA950/GMA965

Date: Tue, 31 Mar 2009 15:31:22 +1100

On Mon, Mar 09, 2009 at 11:50:27PM +0100, Julien BLACHE wrote:
>reassign 518750 libpci3 1:3.1.2-1
>thanks
>
>Filippo Giunchedi <filippo@debian.org> wrote:
>
>Hi Anibal,
>
>See the bug log for the details, summary below.
>
>>mmap(NULL, 1048576, PROT_READ|PROT_WRITE, MAP_SHARED, 4, 0xd0100004) = -1 EINVAL (Invalid argument)
>
>This is a mmap call issued by pommed, with the address taken verbatim
>from libpci.
>
>As you can see in the lspci output, the address is actually
>0xd010000*0*; the 4 we see here is probably the flag indicating a 64bit
>BAR that is not cleared by libpci.
>
>libpci3 from Lenny does pass 0xd0100000 as expected. It looks like a
>subtle bug in libpci; can you investigate this issue and confirm?
>
>Thanks,
>
>JB.
>
>-- 
> Julien BLACHE - Debian & GNU/Linux Developer - <jblache@debian.org> 
> 
> Public key available on <http://www.jblache.org> - KeyID: F5D6 5169 
> GPG Fingerprint : 935A 79F1 C8B3 3521 FD62 7CC7 CD61 4FD7 F5D6 5169 

Hello Martin,

This bug report is available at http://bugs.debian.org/518750

This bug is not present in version 3.0.3 (Lenny), it's in 3.1.2 as
explained above by Julien.

Anibal

Message #107 received at 518750@bugs.debian.org (full text, mbox, reply):

From: Martin Mares <mj@ucw.cz>

To: Aníbal Monsalve Salazar <anibal@debian.org>

Cc: 518750@bugs.debian.org

Subject: Re: Bug#518750: pommed: cannot mmap memory for GMA950/GMA965

Date: Mon, 13 Apr 2009 20:40:13 +0200

> This bug is not present in version 3.0.3 (Lenny), it's in 3.1.2 as
> explained above by Julien.

This is not a bug in libpci, but rather a deficiency in its documentation
leading to improper usage in certain applications.

The pci_dev->base_addr[] array was always intended to contain the BAR
values including the flags in lower bits and it behaved that way for
a long time, except for the sysfs back-end which was setting them
improperly. This bug was however fixed in pciutils-3.1.0
(commit 6d143c3283855c474445a3cf27c65280ed7ab1b7), exposing the
problem in applications to a much larger audience.

I have added a note to the declaration in lib/pci.h explaining
the proper semantics.

				Have a nice fortnight
-- 
Martin `MJ' Mares                          <mj@ucw.cz>   http://mj.ucw.cz/
Faculty of Math and Physics, Charles University, Prague, Czech Rep., Earth
"Never send to know for whom the bell tolls: it tolls for thee." -- John Donne

Message #112 received at 518750@bugs.debian.org (full text, mbox, reply):

From: Guillem Jover <guillem@debian.org>

To: Martin Mares <mj@ucw.cz>, 518750@bugs.debian.org

Subject: Re: Bug#518750: pommed: cannot mmap memory for GMA950/GMA965

Date: Wed, 10 Apr 2019 01:19:51 +0200

Control: reassign -1 pommed

Hi!

On Mon, 2009-04-13 at 20:40:13 +0200, Martin Mares wrote:
> > This bug is not present in version 3.0.3 (Lenny), it's in 3.1.2 as
> > explained above by Julien.
> 
> This is not a bug in libpci, but rather a deficiency in its documentation
> leading to improper usage in certain applications.
> 
> The pci_dev->base_addr[] array was always intended to contain the BAR
> values including the flags in lower bits and it behaved that way for
> a long time, except for the sysfs back-end which was setting them
> improperly. This bug was however fixed in pciutils-3.1.0
> (commit 6d143c3283855c474445a3cf27c65280ed7ab1b7), exposing the
> problem in applications to a much larger audience.
> 
> I have added a note to the declaration in lib/pci.h explaining
> the proper semantics.

So this is really an issue in pommed, as per the pciutils upstream
developer's comment above. Reassigning back.

Thanks,
Guillem

Send a report that this bug log contains spam.