Yo,
Is anyone else seeing instability in recent X600 builds in Fedora (when using 3D desktop effects)? I'm getting occasional X server crashes that never used to happen with this driver - I haven't had chance to track down when it started happening in as of yet but the box stability is restored by disabling desktop effects. I need this box, but I'll schedule some crash time over the weekend or something.
Next time it happens, I'll capture the X server output. There's no oops but a reproducer seems to be along the lines of:
* Login to desktop * Restart VPN and connect to remove IRC proxy * xchat goes nuts with popups catching up replay traffic * kernel ringbuffer shows random segfaults from e.g. DBUS
Not much to go on yet. Just thought I'd share :)
Jon.
On Mon, 2008-10-13 at 22:00 -0400, Jon Masters wrote:
Yo,
Is anyone else seeing instability in recent X600 builds in Fedora (when using 3D desktop effects)? I'm getting occasional X server crashes that never used to happen with this driver - I haven't had chance to track down when it started happening in as of yet but the box stability is restored by disabling desktop effects. I need this box, but I'll schedule some crash time over the weekend or something.
Next time it happens, I'll capture the X server output. There's no oops but a reproducer seems to be along the lines of:
- Login to desktop
- Restart VPN and connect to remove IRC proxy
- xchat goes nuts with popups catching up replay traffic
- kernel ringbuffer shows random segfaults from e.g. DBUS
No useful information, but I (finally after about 8 hours of banging my head against a wall) managed to get F10 installed on this laptop and fully updated to rawhide over the weekend. After about 45 minutes of use X froze hard. No mouse, not keyboard, no nothin'. I didn't turn on anything but defaults on the Beta installer and yum update.
I gave up and booted back into F9 since I do all my work on this thing. But if there is someway I might be able to collect something useful let me know!
01:00.0 VGA compatible controller: ATI Technologies Inc M24 1P [Radeon Mobility X600]
-Eric
On Mon, 2008-10-13 at 22:00 -0400, Jon Masters wrote:
Yo,
Is anyone else seeing instability in recent X600 builds in Fedora (when using 3D desktop effects)? I'm getting occasional X server crashes that never used to happen with this driver - I haven't had chance to track down when it started happening in as of yet but the box stability is restored by disabling desktop effects. I need this box, but I'll schedule some crash time over the weekend or something.
Next time it happens, I'll capture the X server output. There's no oops but a reproducer seems to be along the lines of:
- Login to desktop
- Restart VPN and connect to remove IRC proxy
- xchat goes nuts with popups catching up replay traffic
- kernel ringbuffer shows random segfaults from e.g. DBUS
Not much to go on yet. Just thought I'd share :)
Try with nomodeset maybe but I don't think it'll help.
I'm trying to stabilise the 3D driver again for compiz, we've rewritten radeon support pretty much from scratch for F10 so I'm now fixing the bugs.
Isn't desktop effect unuseably slow for you? I thought the speed might stop people from using it long before the crashing.
Dave.
On Tue, 2008-10-14 at 12:55 +1000, Dave Airlie wrote:
I'm trying to stabilise the 3D driver again for compiz, we've rewritten radeon support pretty much from scratch for F10 so I'm now fixing the bugs.
It's worth saying, this is an F-9 install, with the latest kernel. I do run rawhide, but only under virtual machines with emulated hardware.
Isn't desktop effect unuseably slow for you? I thought the speed might stop people from using it long before the crashing.
Nope, actually for over a year X600 was perfectly rock solid, desktop effects worked just fantastic, and I was actually able to make people who looked at my machine think "hey, this desktop Linux thing aint bad" (ok, they might just think "this Linux thing" there).
So whatever happened in recent *F-9* updates broke it :)
Jon.
On Tue, 2008-10-14 at 00:34 -0400, Jon Masters wrote:
On Tue, 2008-10-14 at 12:55 +1000, Dave Airlie wrote:
I'm trying to stabilise the 3D driver again for compiz, we've rewritten radeon support pretty much from scratch for F10 so I'm now fixing the bugs.
It's worth saying, this is an F-9 install, with the latest kernel. I do run rawhide, but only under virtual machines with emulated hardware.
Isn't desktop effect unuseably slow for you? I thought the speed might stop people from using it long before the crashing.
Nope, actually for over a year X600 was perfectly rock solid, desktop effects worked just fantastic, and I was actually able to make people who looked at my machine think "hey, this desktop Linux thing aint bad" (ok, they might just think "this Linux thing" there).
So whatever happened in recent *F-9* updates broke it :)
Oh that isn't good, I didn't think I'd done much in recent F9 updates.
Try pulling back to the older F9 xorg-x11-drv-ati, say -15.
If its the kernel it must be something I've upstreamed.
Dave.
On Tue, 2008-10-14 at 15:09 +1000, Dave Airlie wrote:
On Tue, 2008-10-14 at 00:34 -0400, Jon Masters wrote:
On Tue, 2008-10-14 at 12:55 +1000, Dave Airlie wrote:
Nope, actually for over a year X600 was perfectly rock solid, desktop effects worked just fantastic, and I was actually able to make people who looked at my machine think "hey, this desktop Linux thing aint bad" (ok, they might just think "this Linux thing" there).
So whatever happened in recent *F-9* updates broke it :)
Oh that isn't good, I didn't think I'd done much in recent F9 updates.
Yeah, that's my concern.
Try pulling back to the older F9 xorg-x11-drv-ati, say -15.
Ok, I'll try that sometime this week - can I do that independently of any other downgrades? (last time I played with X drivers much I was rebuilding the whole X server anyway - my instinct tells me I can just downgrade this package without having problems in this case).
If its the kernel it must be something I've upstreamed.
Yeah, I'll start with the X driver itself. Hopefully it's just something in there that's changed to cause my recent woes.
Jon.
On Tue, 2008-10-14 at 01:20 -0400, Jon Masters wrote:
On Tue, 2008-10-14 at 15:09 +1000, Dave Airlie wrote:
If its the kernel it must be something I've upstreamed.
Yeah, I'll start with the X driver itself. Hopefully it's just something in there that's changed to cause my recent woes.
Of course, I say this and then shortly thereafter Xorg starts chewing 100% CPU and becomes completely wedged. So I login remotely and kill it, which then results in this nice little oops from the radeon driver. I've no time to look at the moment and I need to keep this box more or less running, but stability seems to have gone completely down the pan :(
BUG: unable to handle kernel NULL pointer dereference at 0000000000000000 IP: [<ffffffffa03728a8>] :radeon:radeon_do_cp_idle+0x169/0x1af PGD 11ac1067 PUD e67b067 PMD e687067 PTE 0 Oops: 0000 [1] SMP CPU 0 Modules linked in: nls_utf8 tun vfat fat ipt_MASQUERADE iptable_nat nf_nat nf_conntrack_ipv4 xt_state nf_conntrack ipt_REJECT xt_tcpudp iptable_filter ip_tables x_tables nfsd lockd nfs_acl auth_rpcgss exportfs bridge bnep rfcomm l2cap bluetooth ib_iser rdma_cm ib_cm iw_cm ib_sa ib_mad ib_core ib_addr iscsi_tcp libiscsi scsi_transport_iscsi fuse sunrpc cpufreq_ondemand powernow_k8 freq_table dm_multipath radeon drm ipv6 kvm_amd kvm ppdev sr_mod cdrom snd_hda_intel snd_seq_dummy snd_seq_oss snd_seq_midi_event snd_seq snd_seq_device snd_pcm_oss snd_mixer_oss parport_pc parport pcspkr serio_raw floppy snd_pcm k8temp hwmon r8169 snd_timer firewire_ohci sg firewire_core snd_page_alloc snd_hwdep pata_amd crc_itu_t snd soundcore usb_storage usblp i2c_nforce2 i2c_core dm_snapshot dm_zero dm_mirror dm_log dm_mod ata_generic pata_acpi sata_nv libata sd_mod scsi_mod raid456 async_xor async_memcpy async_tx xor raid1 ext3 jbd mbcache uhci_hcd ohci_hcd ehci_hcd [last unloaded: scsi_wait_scan] Pid: 4102, comm: Xorg Not tainted 2.6.26.5-45.fc9.x86_64 #1 RIP: 0010:[<ffffffffa03728a8>] [<ffffffffa03728a8>] :radeon:radeon_do_cp_idle+0x169/0x1af RSP: 0018:ffff81006d419b18 EFLAGS: 00010202 RAX: 0000000000000000 RBX: ffff81007c151000 RCX: 0000000000026989 RDX: 0000000000026989 RSI: 000000000003ffff RDI: ffff81007c151000 RBP: ffff81006d419b28 R08: ffff81006d418000 R09: 0000000000000000 R10: 0000021906f6f16e R11: 000000007fbebde0 R12: ffff81007da17000 R13: ffff81006780fd80 R14: ffff81006780fdc0 R15: ffff81007da17010 FS: 00007f9763880780(0000) GS:ffffffff81417000(0000) knlGS:00000000ed0e0b90 CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b CR2: 0000000000000000 CR3: 000000000e64d000 CR4: 00000000000006e0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 Process Xorg (pid: 4102, threadinfo ffff81006d418000, task ffff81006413c3e0) Stack: ffff81006780fdc0 ffff81007c151000 ffff81006d419b48 ffffffffa037353b ffff81007da1700c ffff81007da17000 ffff81006d419b58 ffffffffa037cfc9 ffff81006d419b98 ffffffffa0345c96 ffff81006d419b98 ffff81007da1700c Call Trace: [<ffffffffa037353b>] :radeon:radeon_do_release+0x4f/0x12a [<ffffffffa037cfc9>] :radeon:radeon_driver_lastclose+0x9/0xb [<ffffffffa0345c96>] :drm:drm_lastclose+0x61/0x2ce [<ffffffffa03465a3>] :drm:drm_release+0x475/0x492 [<ffffffff810ad44f>] __fput+0xca/0x189 [<ffffffff810ad522>] fput+0x14/0x16 [<ffffffff810aa760>] filp_close+0x66/0x71 [<ffffffff81038c13>] put_files_struct+0x74/0xc8 [<ffffffff81038cae>] exit_files+0x47/0x4f [<ffffffff8103a505>] do_exit+0x293/0x84c [<ffffffff8103ab37>] do_group_exit+0x79/0xa6 [<ffffffff810437b3>] get_signal_to_deliver+0x2a7/0x2cf [<ffffffff8100b363>] do_notify_resume+0x90/0x90c [<ffffffff8104cf3b>] ? ktime_get_ts+0x49/0x4e [<ffffffff8104cf51>] ? ktime_get+0x11/0x42 [<ffffffff8100a75b>] ? __switch_to+0xf5/0x39a [<ffffffff8102a1b4>] ? hrtick_start_fair+0x14a/0x190 [<ffffffff81071f22>] ? audit_syscall_exit+0x331/0x353 [<ffffffff8100c367>] int_signal+0x12/0x17
Code: a0 48 c7 c7 b3 53 38 a0 31 c0 e8 1b 7c f2 e0 eb 03 89 53 28 0f ae f0 83 bb 88 00 00 00 00 74 0f 48 8b 83 20 01 00 00 48 8b 40 18 <8b> 00 eb 11 48 8b 83 00 04 00 00 48 8b 40 18 8b 80 10 07 00 00 RIP [<ffffffffa03728a8>] :radeon:radeon_do_cp_idle+0x169/0x1af RSP <ffff81006d419b18> CR2: 0000000000000000 ---[ end trace 8fe985cc75ef9958 ]--- Fixing recursive fault but reboot is needed!
Jon.
On Tue, 2008-10-14 at 01:47 -0400, Jon Masters wrote:
On Tue, 2008-10-14 at 01:20 -0400, Jon Masters wrote:
On Tue, 2008-10-14 at 15:09 +1000, Dave Airlie wrote:
If its the kernel it must be something I've upstreamed.
Yeah, I'll start with the X driver itself. Hopefully it's just something in there that's changed to cause my recent woes.
Of course, I say this and then shortly thereafter Xorg starts chewing 100% CPU and becomes completely wedged. So I login remotely and kill it, which then results in this nice little oops from the radeon driver. I've no time to look at the moment and I need to keep this box more or less running, but stability seems to have gone completely down the pan :(
The system seems to have been stable since I disabled effects on this driver, which suggests there's a serious regression.
Jon.
kernel@lists.fedoraproject.org
-
David Airlie -
Eric Paris -
Jon Masters -
Jon Masters