[RV630-XT] (Needs 8.602) Memory leak in fglrx

Bug #368271 reported by Evan Huus
12
This bug affects 2 people
Affects Status Importance Assigned to Milestone
fglrx-installer (Ubuntu)
Fix Released
Medium
Unassigned

Bug Description

Binary package hint: xorg-driver-fglrx

Up-to-date Jaunty Final, fresh install. Radeon 2600 XT.

Installing the proprietary drivers works fine, however causes X to leak memory like mad. After about 1h use on fglrx, X was at ~500MB of ram.

Using the open-source driver fixes this problem, but loses 3D acceleration.

I am currently installing the more recent fglrx driver as per http://wiki.cchtml.com/index.php/9.4
I will post back as to whether this fixes the memory leak.

[lspci]
00:00.0 Host bridge [0600]: Intel Corporation 82G33/G31/P35/P31 Express DRAM Controller [8086:29c0] (rev 02)
     Subsystem: Dell Device [1028:020d]
01:00.0 VGA compatible controller [0300]: ATI Technologies Inc RV630 [Radeon HD 2600XT] [1002:9588]
     Subsystem: Dell Device [1028:2542]

Revision history for this message
Evan Huus (eapache) wrote :

The new fglrx drivers as installed at http://wiki.cchtml.com/index.php/9.4
seem to fix the memory leak. After 3h normal usage, X is at 175MB of ram.

I know that the new fglrx doesn't support cards before the HD2000 series, which is why it wasn't included in Jaunty in the first place, however this is a killer memory leak. Is there a way to talk to upstream and get a patch for this issue at the very least?

Or, provide both sets of drivers and allow jockey to differentiate between cards needing the old set, and cards that can use the new set?

Revision history for this message
Bryce Harrington (bryce) wrote :

Hi eapache,

Please attach the output of `lspci -vvnn`, and attach your /var/log/Xorg.0.log (and maybe Xorg.0.log.old) file from after reproducing this issue. If you've made any customizations to your /etc/X11/xorg.conf please attach that as well.

[This is an automated message. Apologies if it has reached you inappropriately; please just reply to this message indicating so.]

tags: added: needs-xorglog
tags: added: needs-lspci-vvnn
Changed in fglrx-installer (Ubuntu):
status: New → Incomplete
Revision history for this message
Evan Huus (eapache) wrote :

I don't think either of those would be useful in this case. It's definitely a driver bug, and appears to be already fixed upstream. The problem is finding a way to fix it that is compatible with SRU rules.

I have attached my lspci output for what it's worth.

Bryce Harrington (bryce)
description: updated
Revision history for this message
Sergio Tosti (zeno979) wrote :

I'm also experiencing this bug, so will put my lspci -vvnn and xorglog in attachment.

Revision history for this message
Sergio Tosti (zeno979) wrote :
Changed in fglrx-installer (Ubuntu):
status: Incomplete → New
Revision history for this message
Sergio Tosti (zeno979) wrote :

Newest version of the binary driver provided by AMD (8.602) isn't affected by bug, I believe that is necessary to update to this version.

Bryce Harrington (bryce)
Changed in fglrx-installer (Ubuntu):
status: New → Confirmed
Bryce Harrington (bryce)
Changed in fglrx-installer (Ubuntu):
importance: Undecided → Medium
status: Confirmed → Triaged
summary: - Memory leak in fglrx
+ (Needs 8.602) Memory leak in fglrx
Bryce Harrington (bryce)
summary: - (Needs 8.602) Memory leak in fglrx
+ [RV630-XT] (Needs 8.602) Memory leak in fglrx
Revision history for this message
Bryce Harrington (bryce) wrote :

I've posted a new version of the -fglrx driver to our xorg-edgers PPA,
would you mind testing it either on Jaunty or Karmic and see if it
resolves this bug?

Get fglrx 8.620 here:

  https://edge.launchpad.net/~xorg-edgers/+archive/ppa

Changed in fglrx-installer (Ubuntu):
status: Triaged → New
status: New → Incomplete
Revision history for this message
Evan Huus (eapache) wrote :

The packages in the ppa do not appear to be affected by this bug. I haven't spent a huge amount of time with them yet, but they should be OK. I'll report back in a few days when I have some serious uptime under my belt.

Revision history for this message
Bryce Harrington (bryce) wrote :

Thanks; yeah the 8.620 changelog specifically mentioned a memory leak on RV630 as fixed, so I'll bet they found and fixed the bug causing this issue. Will look forward to hearing your test results, and we can close this bug when I upload the driver to karmic.

Changed in fglrx-installer (Ubuntu):
status: Incomplete → In Progress
Revision history for this message
Launchpad Janitor (janitor) wrote :

This bug was fixed in the package fglrx-installer - 2:8.620-0ubuntu2

---------------
fglrx-installer (2:8.620-0ubuntu2) karmic; urgency=low

  * Issues resolved since 9.4 (No package changes; Mario just beat me to the upload):
    + With desktop effects enabled, transparent desktop no longer causes desktop icons and
      application windows to disappear
    + Video playback no longer exhibits corruption when desktop effects are enabled and
      Composite is not explicitly disabled
    + [Ubuntu 8.10] System now functions properly when starting X on surround view
      systems
    + Resizing DVD playback window on Mplayer no longer causes blank video
    + System instability no longer occurring when switching to full screen DVD playback
      with Totem Player
    + Terminal window no longer displays error message when launching Catalyst Control
      Center in MultiView
    + [Ubuntu 9.04] Xserver now starts properly with ATI Radeon HD 4870 GPU
      configurations
    + [Ubuntu 9.04] When xrandr --prop is evoked, Xserver failure no longer occurs under
      various system configurations
    + The engine clock and memory clock values are now reported properly by ATI CAL
    + Catalyst Control Center, the primary display is now identified when using the
      Identify Displays button
    + Error message no longer occurs while setting the TV geometry
    + Game corruption does not happen in full screen when 2 monitors are connected and
      RandR 1.2 is enabled
    + TV screen corruption is no longer visible while starting the X-server on some ASICs
    + Executing the following command : aticonfig --initial -f does not cause the following
      message "No supported adapters detected" on some display adapters
    + Detecting second display using amdccle "Detect Display" button no longer results in
      error and will require a relaunch to detect both displays properly
      (LP: #348400)
    + Mplayer does not fail to render video when resizing DVD playback window
    + Fixed: Some system may become unresponsive after starting and killing X-server
      (LP: #355821)
    + System no longer fail to respond when executing multiple server
      generations
    + Memory leak on RV630 (LP: #368271)
    + dpms switches off backlight on laptop screen properly (LP: #363059)
    + Display Preferences no longer causes 100% CPU usage and failure (LP: #366757)

 -- Bryce Harrington <email address hidden> Fri, 26 Jun 2009 22:07:19 -0700

Changed in fglrx-installer (Ubuntu):
status: In Progress → Fix Released
Revision history for this message
Evan Huus (eapache) wrote :

It is definitely fixed. Any chance of this getting an SRU in Jaunty?

Revision history for this message
Bryce Harrington (bryce) wrote : Re: [Bug 368271] Re: [RV630-XT] (Needs 8.602) Memory leak in fglrx

On Sat, Jun 27, 2009 at 12:34:23PM -0000, eapache wrote:
> It is definitely fixed. Any chance of this getting an SRU in Jaunty?

No chance at all, not for proprietary drivers.

But I've stuck it in the x-updates PPA which can be used for updating
drivers for past Ubuntu releases.

Revision history for this message
Evan Huus (eapache) wrote :

Thank you, the x-updates ppa is a little more comforting to use than xorg-edgers :)

Revision history for this message
Sergey Sventitski (sergey-sventitski-deactivatedaccount-deactivatedaccount) wrote :

Installed latest fglrx from ati.com like described here http://wiki.cchtml.com/index.php/Ubuntu_Jaunty_Installation_Guide#Installing_the_drivers_manually and no memory leak problem anymore

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.