Announcement

Collapse
No announcement yet.

AMD Catalyst 13.8 Beta Driver For Linux Released

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #41
    I'm reporting that with Ubuntu 13.04 & Radeon HD 7950 & xorg-edgers ppa (used this to install catalyst 13.8), all Source games are playable and there is no lag, besides delayed audio (presumably caused by pulseaudio, if I can figure out how to force these games to run with pasuspender). FPS is on average ~100 in all Source games, upping to 200-300 FPS in close spaces, and 60FPS when a lot of players are huddled together in the same scene in TF2. There was a bit of frame stutter here and there and steam overlay kills framerates still, but going from 1FPS and being outright unplayable to a smooth-ish 100FPS which is playable is a huge leap. I wonder how long until it can defeat the Windows driver which gets 350-450FPS in all scenarios with maxed AF/AA and a 32 player server.

    Comment


    • #42
      Originally posted by entropy View Post
      As I wrote, I'm currently running this OGL 4.3 beta driver released last week.
      It turns out that a bug I filed http://ati.cchtml.com/show_bug.cgi?id=733
      was fortunately fixed with 13.6b but is "back" again with this OGL 4.3 beta driver.
      So it stands to reason that this driver is based on an older branch than even the 13.6b driver.
      Thank you.

      Comment


      • #43
        I only want improvements on kwin and compiz with catalyst.., I want improvements with xv that continue so laggy...:/..., after 8 months without amd..., all continue on the same way...

        Comment


        • #44
          Use opengl video output or enable tear free.

          Comment


          • #45
            Ok, so i used flim's suggestion and it worked. I could install the driver. Thank you.

            My hardware: A8-5500 APU, 4 GB RAM, Gigabyte F2A85X-D3H mobo
            OS: Debian Testing 64-bit, kernel 3.10.4 (self compiled).

            Now, the bad stuff:

            1. Steam/hl2 based games such as TF2 do not work - something weird happens:

            - The game menus load no problem, then i join a server, the game starts loading and the hl2_linux process starts to grow in memory - it reaches 1.1 GB
            - The join progress bar starts displaying the "sending client info"
            - Then the OS starts swapping everything (and more!) and all processes INCLUDING hl2_linux start being SWAPPED OUT and the system starts to crawl, mouse skipping etc. hl2_linux reaches ~200MB, there is little available memory despite not having any other big programs running and i have 1.7 GB or so swapped out (i dont know from where since there wasnt anything big running, probably hl2 was swapped out too).

            2. Xbmc (which uses opengl and xvba) does something alike - i start a movie, it goes well and after a few minutes it starts to swap out, the movie starts to be choppy etc. Nothing obvious in the memory, but the system still swaps out everything.

            If i kill -9 the offending process (and quit steam too if it was a Steam game) i still dont have that memory back, the system swaps out pretty much anything and slows to a crawl if i want to launch stuff, the used memory is permanently 3.3 GB or so with ~100MB free, i have to reboot.

            After i killed xbmc i had these in dmesg (kernel oops):

            Code:
            [ 1497.630200] INFO: rcu_sched self-detected stall on CPU { 3}  (t=21000 jiffies g=44060 c=44059 q=4640)
            [ 1497.630206] sending NMI to all CPUs:
            [ 1497.630212] NMI backtrace for cpu 3
            [ 1497.630216] CPU: 3 PID: 5701 Comm: xbmc.bin Tainted: P           O 3.10.4 #1
            [ 1497.630218] Hardware name: Gigabyte Technology Co., Ltd. To be filled by O.E.M./F2A85X-D3H, BIOS F1 10/09/2012
            [ 1497.630221] task: ffff880135d92140 ti: ffff88001ef46000 task.ti: ffff88001ef46000
            [ 1497.630223] RIP: 0010:[<ffffffff811a664b>]  [<ffffffff811a664b>] __const_udelay+0x9/0x23
            [ 1497.630229] RSP: 0018:ffff88013ed83e60  EFLAGS: 00000046
            [ 1497.630232] RAX: 0000000000000000 RBX: 0000000000002710 RCX: 0000000000000040
            [ 1497.630234] RDX: 000000000030bcaf RSI: 0000000000000200 RDI: 0000000000418958
            [ 1497.630236] RBP: ffff88013ed8d9a0 R08: 0000000000000002 R09: ffffffff814e3b9d
            [ 1497.630238] R10: 0000000000000000 R11: ffff880107b5a200 R12: ffffffff81637e40
            [ 1497.630240] R13: ffff88013ed8d370 R14: ffff88001ef46000 R15: 0000000000000003
            [ 1497.630242] FS:  00007f4f69de4700(0000) GS:ffff88013ed80000(0000) knlGS:00000000daafeb40
            [ 1497.630244] CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
            [ 1497.630246] CR2: 00007f376d304228 CR3: 0000000114932000 CR4: 00000000000407e0
            [ 1497.630247] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
            [ 1497.630249] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
            [ 1497.630250] Stack:
            [ 1497.630252]  ffffffff8101c67f ffffffff81637e40 ffffffff81092a8c 0000000000001220
            [ 1497.630256]  0000000000076ad7 0000000000000003 ffff880135d92140 0000000000000000
            [ 1497.630259]  0000000000000003 ffff88013ed8d370 ffffffff8106bbb2 ffff88013ed83f60
            [ 1497.630262] Call Trace:
            [ 1497.630264]  <IRQ> 
            [ 1497.630265]  [<ffffffff8101c67f>] ? arch_trigger_all_cpu_backtrace+0x76/0x7f
            [ 1497.630274]  [<ffffffff81092a8c>] ? rcu_check_callbacks+0x1a4/0x4bb
            [ 1497.630278]  [<ffffffff8106bbb2>] ? tick_sched_do_timer+0x25/0x25
            [ 1497.630282]  [<ffffffff8103ab79>] ? update_process_times+0x31/0x5c
            [ 1497.630285]  [<ffffffff8106b916>] ? tick_sched_handle+0x3e/0x4a
            [ 1497.630288]  [<ffffffff8106bbe2>] ? tick_sched_timer+0x30/0x4c
            [ 1497.630292]  [<ffffffff8104c09c>] ? __run_hrtimer+0xa9/0x14e
            [ 1497.630295]  [<ffffffff8104c6a5>] ? hrtimer_interrupt+0xbd/0x19e
            [ 1497.630299]  [<ffffffff8101bc60>] ? smp_apic_timer_interrupt+0x6d/0x7e
            [ 1497.630302]  [<ffffffff813624ca>] ? apic_timer_interrupt+0x6a/0x70
            [ 1497.630304]  <EOI> 
            [ 1497.630312]  [<ffffffffa03657dc>] ? drm_free+0xcc/0x1a0 [fglrx]
            [ 1497.630425]  [<ffffffffa0371228>] ? firegl_remove_mutex+0x68/0x180 [fglrx]
            [ 1497.630471]  [<ffffffffa0371209>] ? firegl_remove_mutex+0x49/0x180 [fglrx]
            [ 1497.630515]  [<ffffffffa036b0c1>] ? firegl_release_helper+0x1c1/0x700 [fglrx]
            [ 1497.630557]  [<ffffffffa036d050>] ? firegl_release+0x60/0x1b0 [fglrx]
            [ 1497.630561]  [<ffffffff810fb3a2>] ? __fput+0xe1/0x1b1
            [ 1497.630564]  [<ffffffff8104763a>] ? task_work_run+0x7b/0x91
            [ 1497.630568]  [<ffffffff81032b6e>] ? do_exit+0x3e2/0x90c
            [ 1497.630571]  [<ffffffff81033122>] ? do_group_exit+0x66/0x98
            [ 1497.630574]  [<ffffffff8103dd01>] ? get_signal_to_deliver+0x479/0x4ad
            [ 1497.630577]  [<ffffffff8100205d>] ? do_signal+0x3c/0x432
            [ 1497.630580]  [<ffffffff810fa4ef>] ? vfs_read+0xa9/0xef
            [ 1497.630583]  [<ffffffff810fb83d>] ? fput+0xe/0xb6
            [ 1497.630586]  [<ffffffff81002473>] ? do_notify_resume+0x20/0x5d
            [ 1497.630589]  [<ffffffff81361b9a>] ? int_signal+0x12/0x17
            [ 1497.630590] Code: eb 0e 66 66 66 66 66 2e 0f 1f 84 00 00 00 00 00 48 ff c8 75 fb 48 ff c8 c3 48 8b 05 f8 3b 4b 00 ff e0 65 48 8b 14 25 60 26 01 00 <48> 69 d2 fa 00 00 00 48 8d 04 bd 00 00 00 00 f7 e2 48 8d 7a 01 
            [ 1497.630622] NMI backtrace for cpu 2
            [ 1497.630625] CPU: 2 PID: 0 Comm: swapper/2 Tainted: P           O 3.10.4 #1
            [ 1497.630627] Hardware name: Gigabyte Technology Co., Ltd. To be filled by O.E.M./F2A85X-D3H, BIOS F1 10/09/2012
            [ 1497.630629] task: ffff8801394f1040 ti: ffff88013901a000 task.ti: ffff88013901a000
            [ 1497.630630] RIP: 0010:[<ffffffffa02b5ac8>]  [<ffffffffa02b5ac8>] acpi_idle_do_entry+0x16/0x1f [processor]
            [ 1497.630638] RSP: 0018:ffff88013901be60  EFLAGS: 00000093
            [ 1497.630640] RAX: 0000015cb1a44700 RBX: ffff88013611a4a0 RCX: ffff88013ed00000
            [ 1497.630641] RDX: 0000000000001771 RSI: 000000000000b870 RDI: ffff88013611a4a0
            [ 1497.630643] RBP: 0000000000000002 R08: 0000000000001472 R09: 0000000000000d72
            [ 1497.630644] R10: 0000000000000064 R11: 0000000000001536 R12: ffff88013901a000
            [ 1497.630646] R13: ffff88013611a400 R14: 0000000000000000 R15: 0000000000000000
            [ 1497.630648] FS:  00007f31c7224700(0000) GS:ffff88013ed00000(0000) knlGS:00000000daafeb40
            [ 1497.630650] CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
            [ 1497.630652] CR2: 00007f31c746e9e0 CR3: 0000000111b5c000 CR4: 00000000000407e0
            [ 1497.630653] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
            [ 1497.630655] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
            [ 1497.630656] Stack:
            [ 1497.630657]  ffffffffa02b5bbb ffff880135b58e00 ffff88013901bee0 ffffffffa02b9238
            [ 1497.630661]  0000015c30b4b5a4 ffffffff81276161 0000000000000002 ffffffffa02b9170
            [ 1497.630664]  0000000000000000 00000000004fdf1e ffff88013ed0d610 ffff880135b58e00
            [ 1497.630667] Call Trace:
            [ 1497.630673]  [<ffffffffa02b5bbb>] ? acpi_idle_enter_simple+0x89/0xc3 [processor]
            [ 1497.630678]  [<ffffffff81276161>] ? cpuidle_enter_state+0x40/0xb6
            [ 1497.630684]  [<ffffffff812762a7>] ? cpuidle_idle_call+0xd0/0x13f
            [ 1497.630688]  [<ffffffff810086ec>] ? arch_cpu_idle+0x5/0x14
            [ 1497.630692]  [<ffffffff81064881>] ? cpu_startup_entry+0x103/0x176
            [ 1497.630696]  [<ffffffff8134eba9>] ? start_secondary+0x1d9/0x1dd
            [ 1497.630697] Code: 03 fb f4 fa 65 48 8b 04 25 60 b9 00 00 83 88 3c e0 ff ff 04 c3 8a 47 08 3c 01 75 05 e9 33 52 d6 e0 3c 02 75 02 eb be 8b 57 04 ec <48> 8b 15 05 d7 57 e1 ed c3 41 54 48 63 c2 48 8d 04 c5 e0 3a 01 
            [ 1497.630738] NMI backtrace for cpu 0
            [ 1497.630751] CPU: 0 PID: 0 Comm: swapper/0 Tainted: P           O 3.10.4 #1
            [ 1497.630759] Hardware name: Gigabyte Technology Co., Ltd. To be filled by O.E.M./F2A85X-D3H, BIOS F1 10/09/2012
            [ 1497.630765] task: ffffffff81610400 ti: ffffffff81600000 task.ti: ffffffff81600000
            [ 1497.630770] RIP: 0010:[<ffffffffa02b5aa0>]  [<ffffffffa02b5aa0>] acpi_safe_halt+0x1e/0x30 [processor]
            [ 1497.630797] RSP: 0018:ffffffff81601ee8  EFLAGS: 00000246
            [ 1497.630802] RAX: 0000000000000000 RBX: ffff880135d5d86c RCX: 0000000000000018
            [ 1497.630806] RDX: 0000000000000001 RSI: 0000000000000001 RDI: ffff880135d5d86c
            [ 1497.630810] RBP: 0000000000000001 R08: 000000000000038b R09: 000000000000038a
            [ 1497.630815] R10: 00000000000013ec R11: 000000000000131e R12: ffff880135d5d800
            [ 1497.630819] R13: 0000015c30b9d09e R14: 0000000000000000 R15: 000000009bf2e018
            [ 1497.630825] FS:  00007ffaa7ff5740(0000) GS:ffff88013ec00000(0000) knlGS:00000000f5936980
            [ 1497.630829] CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
            [ 1497.630833] CR2: 00000000b774efac CR3: 0000000111891000 CR4: 00000000000407f0
            [ 1497.630837] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
            [ 1497.630842] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
            [ 1497.630845] Stack:
            [ 1497.630848]  ffffffffa02b5b17 ffff88013556d200 ffffffff81601f60 ffffffffa02b91e0
            [ 1497.630858]  ffffffff81276161 0000000000000001 ffffffffa02b9170 0000000000000000
            [ 1497.630867]  00000000000dd703 ffff88013ec0d810 ffff88013556d200 ffffffffa02b9170
            [ 1497.630875] Call Trace:
            [ 1497.630891]  [<ffffffffa02b5b17>] ? acpi_idle_enter_c1+0x46/0x61 [processor]
            [ 1497.630905]  [<ffffffff81276161>] ? cpuidle_enter_state+0x40/0xb6
            [ 1497.630920]  [<ffffffff812762a7>] ? cpuidle_idle_call+0xd0/0x13f
            [ 1497.630930]  [<ffffffff810086ec>] ? arch_cpu_idle+0x5/0x14
            [ 1497.630938]  [<ffffffff81064881>] ? cpu_startup_entry+0x103/0x176
            [ 1497.630947]  [<ffffffff816a7cb4>] ? start_kernel+0x3d6/0x3e1
            [ 1497.630954]  [<ffffffff816a7700>] ? repair_env_string+0x54/0x54
            [ 1497.630958] Code: 80 38 e0 ff ff 48 c1 e8 03 83 e0 01 c3 65 48 8b 04 25 60 b9 00 00 83 a0 3c e0 ff ff fb 0f ae f0 e8 d0 ff ff ff 85 c0 75 03 fb f4 <fa> 65 48 8b 04 25 60 b9 00 00 83 88 3c e0 ff ff 04 c3 8a 47 08 
            [ 1497.631035] NMI backtrace for cpu 1
            [ 1497.631042] CPU: 1 PID: 0 Comm: swapper/1 Tainted: P           O 3.10.4 #1
            [ 1497.631045] Hardware name: Gigabyte Technology Co., Ltd. To be filled by O.E.M./F2A85X-D3H, BIOS F1 10/09/2012
            [ 1497.631049] task: ffff8801394f1770 ti: ffff880139016000 task.ti: ffff880139016000
            [ 1497.631052] RIP: 0010:[<ffffffffa02b5ac8>]  [<ffffffffa02b5ac8>] acpi_idle_do_entry+0x16/0x1f [processor]
            [ 1497.631065] RSP: 0018:ffff880139017e60  EFLAGS: 00000093
            [ 1497.631068] RAX: 0000015cb1a5a400 RBX: ffff880135be8ca0 RCX: ffff88013ec80000
            [ 1497.631072] RDX: 0000000000001771 RSI: 000000000000b870 RDI: ffff880135be8ca0
            [ 1497.631075] RBP: 0000000000000002 R08: 0000000000001419 R09: 0000000000000d02
            [ 1497.631079] R10: 000000000000044c R11: 0000000000000b7b R12: ffff880139016000
            [ 1497.631082] R13: ffff880135be8c00 R14: 0000000000000000 R15: 0000000000000000
            [ 1497.631086] FS:  00007ffaa22d7700(0000) GS:ffff88013ec80000(0000) knlGS:00000000f5936980
            [ 1497.631090] CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
            [ 1497.631093] CR2: 00000000b774efac CR3: 0000000111891000 CR4: 00000000000407e0
            [ 1497.631096] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
            [ 1497.631100] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
            [ 1497.631102] Stack:
            [ 1497.631104]  ffffffffa02b5bbb ffff880135b58200 ffff880139017ee0 ffffffffa02b9238
            [ 1497.631110]  0000015c30b61254 ffffffff81276161 0000000000000002 ffffffffa02b9170
            [ 1497.631117]  0000000000000000 00000000004e8351 ffff88013ec8d610 ffff880135b58200
            [ 1497.631122] Call Trace:
            [ 1497.631135]  [<ffffffffa02b5bbb>] ? acpi_idle_enter_simple+0x89/0xc3 [processor]
            [ 1497.631145]  [<ffffffff81276161>] ? cpuidle_enter_state+0x40/0xb6
            [ 1497.631158]  [<ffffffff812762a7>] ? cpuidle_idle_call+0xd0/0x13f
            [ 1497.631164]  [<ffffffff810086ec>] ? arch_cpu_idle+0x5/0x14
            [ 1497.631170]  [<ffffffff81064881>] ? cpu_startup_entry+0x103/0x176
            [ 1497.631176]  [<ffffffff8134eba9>] ? start_secondary+0x1d9/0x1dd
            [ 1497.631179] Code: 03 fb f4 fa 65 48 8b 04 25 60 b9 00 00 83 88 3c e0 ff ff 04 c3 8a 47 08 3c 01 75 05 e9 33 52 d6 e0 3c 02 75 02 eb be 8b 57 04 ec <48> 8b 15 05 d7 57 e1 ed c3 41 54 48 63 c2 48 8d 04 c5 e0 3a 01
            If i killed the offending process, an instance of "migration" process appears using 90+ % CPU - i dont know what it is exactly, it has 0 at every memory category (in top) - i asume its a kernel process, but after it goes away after a minute or so, no memory seems to be regained or anything.

            I wonder what the hell is going on and it sems a massive memory leak from somewhere or perhaps the driver grabs some memory it shouldnt? Clearly it locks down tons of memory from the system.
            I had the IGP memory set to 512 MB (then to "auto", but it reported the same 512 MB), and indeed i had 3.4 GB usable RAM out of 4 so thats a given, but what the hell is with the rest?

            When booting, i have this in the system log:

            Code:
            [    7.720360] <6>[fglrx] Maximum main memory to use for locked dma buffers: 3194 MBytes.
            I dont remember seeing it with older drivers, i remember seeing a value of 491 (or so) MB, but not sure if it was in this context exactly.

            Othe rgames like hl1 based ones and Urban Terror-likes work well.

            Now, with the previous drivers (up to 13.6 beta) hl2_linux used quite some memory, around 2GB, xbmc used just about as with this one, but i had no issues.
            The previous prerelease driver, 13.15 "OpenGl 4.3" version also had these issuess, but i didnt really bother with it as i switched back to 13.6 and hoped the next will cure them. No such luck.


            Anyone else with these issues?


            PS. I know that the memory in Linux seemss full sometimes but it has buffered/cached stuff in it. This is NOT the case here: If i kill some process, the memory is instantly freed up and it shows up in top. Also, the buffered/cached stuff is moved out of the way if an active process needs memory. Here the memory seems full but no process uses much of it, and the total used memory is definitely not ~3.3 GB.
            Also, this is happening ONLY with these 2 recent beta drivers (13.15 and 13.20)
            Last edited by gradinaruvasile; 02 August 2013, 05:09 PM.

            Comment


            • #46
              Better to report these issues forward to valve and ati.

              Comment


              • #47
                Originally posted by gradinaruvasile View Post
                Ok, so i used flim's suggestion and it worked. I could install the driver. Thank you.

                My hardware: A8-5500 APU, 4 GB RAM, Gigabyte F2A85X-D3H mobo
                OS: Debian Testing 64-bit, kernel 3.10.4 (self compiled).

                Now, the bad stuff:

                1. Steam/hl2 based games such as TF2 do not work - something weird happens:

                - The game menus load no problem, then i join a server, the game starts loading and the hl2_linux process starts to grow in memory - it reaches 1.1 GB
                - The join progress bar starts displaying the "sending client info"
                - Then the OS starts swapping everything (and more!) and all processes INCLUDING hl2_linux start being SWAPPED OUT and the system starts to crawl, mouse skipping etc. hl2_linux reaches ~200MB, there is little available memory despite not having any other big programs running and i have 1.7 GB or so swapped out (i dont know from where since there wasnt anything big running, probably hl2 was swapped out too).

                2. Xbmc (which uses opengl and xvba) does something alike - i start a movie, it goes well and after a few minutes it starts to swap out, the movie starts to be choppy etc. Nothing obvious in the memory, but the system still swaps out everything.

                If i kill -9 the offending process (and quit steam too if it was a Steam game) i still dont have that memory back, the system swaps out pretty much anything and slows to a crawl if i want to launch stuff, the used memory is permanently 3.3 GB or so with ~100MB free, i have to reboot.

                After i killed xbmc i had these in dmesg (kernel oops):

                Code:
                [ 1497.630200] INFO: rcu_sched self-detected stall on CPU { 3}  (t=21000 jiffies g=44060 c=44059 q=4640)
                [ 1497.630206] sending NMI to all CPUs:
                [ 1497.630212] NMI backtrace for cpu 3
                [ 1497.630216] CPU: 3 PID: 5701 Comm: xbmc.bin Tainted: P           O 3.10.4 #1
                [ 1497.630218] Hardware name: Gigabyte Technology Co., Ltd. To be filled by O.E.M./F2A85X-D3H, BIOS F1 10/09/2012
                [ 1497.630221] task: ffff880135d92140 ti: ffff88001ef46000 task.ti: ffff88001ef46000
                [ 1497.630223] RIP: 0010:[<ffffffff811a664b>]  [<ffffffff811a664b>] __const_udelay+0x9/0x23
                [ 1497.630229] RSP: 0018:ffff88013ed83e60  EFLAGS: 00000046
                [ 1497.630232] RAX: 0000000000000000 RBX: 0000000000002710 RCX: 0000000000000040
                [ 1497.630234] RDX: 000000000030bcaf RSI: 0000000000000200 RDI: 0000000000418958
                [ 1497.630236] RBP: ffff88013ed8d9a0 R08: 0000000000000002 R09: ffffffff814e3b9d
                [ 1497.630238] R10: 0000000000000000 R11: ffff880107b5a200 R12: ffffffff81637e40
                [ 1497.630240] R13: ffff88013ed8d370 R14: ffff88001ef46000 R15: 0000000000000003
                [ 1497.630242] FS:  00007f4f69de4700(0000) GS:ffff88013ed80000(0000) knlGS:00000000daafeb40
                [ 1497.630244] CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
                [ 1497.630246] CR2: 00007f376d304228 CR3: 0000000114932000 CR4: 00000000000407e0
                [ 1497.630247] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
                [ 1497.630249] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
                [ 1497.630250] Stack:
                [ 1497.630252]  ffffffff8101c67f ffffffff81637e40 ffffffff81092a8c 0000000000001220
                [ 1497.630256]  0000000000076ad7 0000000000000003 ffff880135d92140 0000000000000000
                [ 1497.630259]  0000000000000003 ffff88013ed8d370 ffffffff8106bbb2 ffff88013ed83f60
                [ 1497.630262] Call Trace:
                [ 1497.630264]  <IRQ> 
                [ 1497.630265]  [<ffffffff8101c67f>] ? arch_trigger_all_cpu_backtrace+0x76/0x7f
                [ 1497.630274]  [<ffffffff81092a8c>] ? rcu_check_callbacks+0x1a4/0x4bb
                [ 1497.630278]  [<ffffffff8106bbb2>] ? tick_sched_do_timer+0x25/0x25
                [ 1497.630282]  [<ffffffff8103ab79>] ? update_process_times+0x31/0x5c
                [ 1497.630285]  [<ffffffff8106b916>] ? tick_sched_handle+0x3e/0x4a
                [ 1497.630288]  [<ffffffff8106bbe2>] ? tick_sched_timer+0x30/0x4c
                [ 1497.630292]  [<ffffffff8104c09c>] ? __run_hrtimer+0xa9/0x14e
                [ 1497.630295]  [<ffffffff8104c6a5>] ? hrtimer_interrupt+0xbd/0x19e
                [ 1497.630299]  [<ffffffff8101bc60>] ? smp_apic_timer_interrupt+0x6d/0x7e
                [ 1497.630302]  [<ffffffff813624ca>] ? apic_timer_interrupt+0x6a/0x70
                [ 1497.630304]  <EOI> 
                [ 1497.630312]  [<ffffffffa03657dc>] ? drm_free+0xcc/0x1a0 [fglrx]
                [ 1497.630425]  [<ffffffffa0371228>] ? firegl_remove_mutex+0x68/0x180 [fglrx]
                [ 1497.630471]  [<ffffffffa0371209>] ? firegl_remove_mutex+0x49/0x180 [fglrx]
                [ 1497.630515]  [<ffffffffa036b0c1>] ? firegl_release_helper+0x1c1/0x700 [fglrx]
                [ 1497.630557]  [<ffffffffa036d050>] ? firegl_release+0x60/0x1b0 [fglrx]
                [ 1497.630561]  [<ffffffff810fb3a2>] ? __fput+0xe1/0x1b1
                [ 1497.630564]  [<ffffffff8104763a>] ? task_work_run+0x7b/0x91
                [ 1497.630568]  [<ffffffff81032b6e>] ? do_exit+0x3e2/0x90c
                [ 1497.630571]  [<ffffffff81033122>] ? do_group_exit+0x66/0x98
                [ 1497.630574]  [<ffffffff8103dd01>] ? get_signal_to_deliver+0x479/0x4ad
                [ 1497.630577]  [<ffffffff8100205d>] ? do_signal+0x3c/0x432
                [ 1497.630580]  [<ffffffff810fa4ef>] ? vfs_read+0xa9/0xef
                [ 1497.630583]  [<ffffffff810fb83d>] ? fput+0xe/0xb6
                [ 1497.630586]  [<ffffffff81002473>] ? do_notify_resume+0x20/0x5d
                [ 1497.630589]  [<ffffffff81361b9a>] ? int_signal+0x12/0x17
                [ 1497.630590] Code: eb 0e 66 66 66 66 66 2e 0f 1f 84 00 00 00 00 00 48 ff c8 75 fb 48 ff c8 c3 48 8b 05 f8 3b 4b 00 ff e0 65 48 8b 14 25 60 26 01 00 <48> 69 d2 fa 00 00 00 48 8d 04 bd 00 00 00 00 f7 e2 48 8d 7a 01 
                [ 1497.630622] NMI backtrace for cpu 2
                [ 1497.630625] CPU: 2 PID: 0 Comm: swapper/2 Tainted: P           O 3.10.4 #1
                [ 1497.630627] Hardware name: Gigabyte Technology Co., Ltd. To be filled by O.E.M./F2A85X-D3H, BIOS F1 10/09/2012
                [ 1497.630629] task: ffff8801394f1040 ti: ffff88013901a000 task.ti: ffff88013901a000
                [ 1497.630630] RIP: 0010:[<ffffffffa02b5ac8>]  [<ffffffffa02b5ac8>] acpi_idle_do_entry+0x16/0x1f [processor]
                [ 1497.630638] RSP: 0018:ffff88013901be60  EFLAGS: 00000093
                [ 1497.630640] RAX: 0000015cb1a44700 RBX: ffff88013611a4a0 RCX: ffff88013ed00000
                [ 1497.630641] RDX: 0000000000001771 RSI: 000000000000b870 RDI: ffff88013611a4a0
                [ 1497.630643] RBP: 0000000000000002 R08: 0000000000001472 R09: 0000000000000d72
                [ 1497.630644] R10: 0000000000000064 R11: 0000000000001536 R12: ffff88013901a000
                [ 1497.630646] R13: ffff88013611a400 R14: 0000000000000000 R15: 0000000000000000
                [ 1497.630648] FS:  00007f31c7224700(0000) GS:ffff88013ed00000(0000) knlGS:00000000daafeb40
                [ 1497.630650] CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
                [ 1497.630652] CR2: 00007f31c746e9e0 CR3: 0000000111b5c000 CR4: 00000000000407e0
                [ 1497.630653] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
                [ 1497.630655] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
                [ 1497.630656] Stack:
                [ 1497.630657]  ffffffffa02b5bbb ffff880135b58e00 ffff88013901bee0 ffffffffa02b9238
                [ 1497.630661]  0000015c30b4b5a4 ffffffff81276161 0000000000000002 ffffffffa02b9170
                [ 1497.630664]  0000000000000000 00000000004fdf1e ffff88013ed0d610 ffff880135b58e00
                [ 1497.630667] Call Trace:
                [ 1497.630673]  [<ffffffffa02b5bbb>] ? acpi_idle_enter_simple+0x89/0xc3 [processor]
                [ 1497.630678]  [<ffffffff81276161>] ? cpuidle_enter_state+0x40/0xb6
                [ 1497.630684]  [<ffffffff812762a7>] ? cpuidle_idle_call+0xd0/0x13f
                [ 1497.630688]  [<ffffffff810086ec>] ? arch_cpu_idle+0x5/0x14
                [ 1497.630692]  [<ffffffff81064881>] ? cpu_startup_entry+0x103/0x176
                [ 1497.630696]  [<ffffffff8134eba9>] ? start_secondary+0x1d9/0x1dd
                [ 1497.630697] Code: 03 fb f4 fa 65 48 8b 04 25 60 b9 00 00 83 88 3c e0 ff ff 04 c3 8a 47 08 3c 01 75 05 e9 33 52 d6 e0 3c 02 75 02 eb be 8b 57 04 ec <48> 8b 15 05 d7 57 e1 ed c3 41 54 48 63 c2 48 8d 04 c5 e0 3a 01 
                [ 1497.630738] NMI backtrace for cpu 0
                [ 1497.630751] CPU: 0 PID: 0 Comm: swapper/0 Tainted: P           O 3.10.4 #1
                [ 1497.630759] Hardware name: Gigabyte Technology Co., Ltd. To be filled by O.E.M./F2A85X-D3H, BIOS F1 10/09/2012
                [ 1497.630765] task: ffffffff81610400 ti: ffffffff81600000 task.ti: ffffffff81600000
                [ 1497.630770] RIP: 0010:[<ffffffffa02b5aa0>]  [<ffffffffa02b5aa0>] acpi_safe_halt+0x1e/0x30 [processor]
                [ 1497.630797] RSP: 0018:ffffffff81601ee8  EFLAGS: 00000246
                [ 1497.630802] RAX: 0000000000000000 RBX: ffff880135d5d86c RCX: 0000000000000018
                [ 1497.630806] RDX: 0000000000000001 RSI: 0000000000000001 RDI: ffff880135d5d86c
                [ 1497.630810] RBP: 0000000000000001 R08: 000000000000038b R09: 000000000000038a
                [ 1497.630815] R10: 00000000000013ec R11: 000000000000131e R12: ffff880135d5d800
                [ 1497.630819] R13: 0000015c30b9d09e R14: 0000000000000000 R15: 000000009bf2e018
                [ 1497.630825] FS:  00007ffaa7ff5740(0000) GS:ffff88013ec00000(0000) knlGS:00000000f5936980
                [ 1497.630829] CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
                [ 1497.630833] CR2: 00000000b774efac CR3: 0000000111891000 CR4: 00000000000407f0
                [ 1497.630837] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
                [ 1497.630842] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
                [ 1497.630845] Stack:
                [ 1497.630848]  ffffffffa02b5b17 ffff88013556d200 ffffffff81601f60 ffffffffa02b91e0
                [ 1497.630858]  ffffffff81276161 0000000000000001 ffffffffa02b9170 0000000000000000
                [ 1497.630867]  00000000000dd703 ffff88013ec0d810 ffff88013556d200 ffffffffa02b9170
                [ 1497.630875] Call Trace:
                [ 1497.630891]  [<ffffffffa02b5b17>] ? acpi_idle_enter_c1+0x46/0x61 [processor]
                [ 1497.630905]  [<ffffffff81276161>] ? cpuidle_enter_state+0x40/0xb6
                [ 1497.630920]  [<ffffffff812762a7>] ? cpuidle_idle_call+0xd0/0x13f
                [ 1497.630930]  [<ffffffff810086ec>] ? arch_cpu_idle+0x5/0x14
                [ 1497.630938]  [<ffffffff81064881>] ? cpu_startup_entry+0x103/0x176
                [ 1497.630947]  [<ffffffff816a7cb4>] ? start_kernel+0x3d6/0x3e1
                [ 1497.630954]  [<ffffffff816a7700>] ? repair_env_string+0x54/0x54
                [ 1497.630958] Code: 80 38 e0 ff ff 48 c1 e8 03 83 e0 01 c3 65 48 8b 04 25 60 b9 00 00 83 a0 3c e0 ff ff fb 0f ae f0 e8 d0 ff ff ff 85 c0 75 03 fb f4 <fa> 65 48 8b 04 25 60 b9 00 00 83 88 3c e0 ff ff 04 c3 8a 47 08 
                [ 1497.631035] NMI backtrace for cpu 1
                [ 1497.631042] CPU: 1 PID: 0 Comm: swapper/1 Tainted: P           O 3.10.4 #1
                [ 1497.631045] Hardware name: Gigabyte Technology Co., Ltd. To be filled by O.E.M./F2A85X-D3H, BIOS F1 10/09/2012
                [ 1497.631049] task: ffff8801394f1770 ti: ffff880139016000 task.ti: ffff880139016000
                [ 1497.631052] RIP: 0010:[<ffffffffa02b5ac8>]  [<ffffffffa02b5ac8>] acpi_idle_do_entry+0x16/0x1f [processor]
                [ 1497.631065] RSP: 0018:ffff880139017e60  EFLAGS: 00000093
                [ 1497.631068] RAX: 0000015cb1a5a400 RBX: ffff880135be8ca0 RCX: ffff88013ec80000
                [ 1497.631072] RDX: 0000000000001771 RSI: 000000000000b870 RDI: ffff880135be8ca0
                [ 1497.631075] RBP: 0000000000000002 R08: 0000000000001419 R09: 0000000000000d02
                [ 1497.631079] R10: 000000000000044c R11: 0000000000000b7b R12: ffff880139016000
                [ 1497.631082] R13: ffff880135be8c00 R14: 0000000000000000 R15: 0000000000000000
                [ 1497.631086] FS:  00007ffaa22d7700(0000) GS:ffff88013ec80000(0000) knlGS:00000000f5936980
                [ 1497.631090] CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
                [ 1497.631093] CR2: 00000000b774efac CR3: 0000000111891000 CR4: 00000000000407e0
                [ 1497.631096] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
                [ 1497.631100] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
                [ 1497.631102] Stack:
                [ 1497.631104]  ffffffffa02b5bbb ffff880135b58200 ffff880139017ee0 ffffffffa02b9238
                [ 1497.631110]  0000015c30b61254 ffffffff81276161 0000000000000002 ffffffffa02b9170
                [ 1497.631117]  0000000000000000 00000000004e8351 ffff88013ec8d610 ffff880135b58200
                [ 1497.631122] Call Trace:
                [ 1497.631135]  [<ffffffffa02b5bbb>] ? acpi_idle_enter_simple+0x89/0xc3 [processor]
                [ 1497.631145]  [<ffffffff81276161>] ? cpuidle_enter_state+0x40/0xb6
                [ 1497.631158]  [<ffffffff812762a7>] ? cpuidle_idle_call+0xd0/0x13f
                [ 1497.631164]  [<ffffffff810086ec>] ? arch_cpu_idle+0x5/0x14
                [ 1497.631170]  [<ffffffff81064881>] ? cpu_startup_entry+0x103/0x176
                [ 1497.631176]  [<ffffffff8134eba9>] ? start_secondary+0x1d9/0x1dd
                [ 1497.631179] Code: 03 fb f4 fa 65 48 8b 04 25 60 b9 00 00 83 88 3c e0 ff ff 04 c3 8a 47 08 3c 01 75 05 e9 33 52 d6 e0 3c 02 75 02 eb be 8b 57 04 ec <48> 8b 15 05 d7 57 e1 ed c3 41 54 48 63 c2 48 8d 04 c5 e0 3a 01
                If i killed the offending process, an instance of "migration" process appears using 90+ % CPU - i dont know what it is exactly, it has 0 at every memory category (in top) - i asume its a kernel process, but after it goes away after a minute or so, no memory seems to be regained or anything.

                I wonder what the hell is going on and it sems a massive memory leak from somewhere or perhaps the driver grabs some memory it shouldnt? Clearly it locks down tons of memory from the system.
                I had the IGP memory set to 512 MB (then to "auto", but it reported the same 512 MB), and indeed i had 3.4 GB usable RAM out of 4 so thats a given, but what the hell is with the rest?

                When booting, i have this in the system log:

                Code:
                [    7.720360] <6>[fglrx] Maximum main memory to use for locked dma buffers: 3194 MBytes.
                I dont remember seeing it with older drivers, i remember seeing a value of 491 (or so) MB, but not sure if it was in this context exactly.

                Othe rgames like hl1 based ones and Urban Terror-likes work well.

                Now, with the previous drivers (up to 13.6 beta) hl2_linux used quite some memory, around 2GB, xbmc used just about as with this one, but i had no issues.
                The previous prerelease driver, 13.15 "OpenGl 4.3" version also had these issuess, but i didnt really bother with it as i switched back to 13.6 and hoped the next will cure them. No such luck.


                Anyone else with these issues?


                PS. I know that the memory in Linux seemss full sometimes but it has buffered/cached stuff in it. This is NOT the case here: If i kill some process, the memory is instantly freed up and it shows up in top. Also, the buffered/cached stuff is moved out of the way if an active process needs memory. Here the memory seems full but no process uses much of it, and the total used memory is definitely not ~3.3 GB.
                Also, this is happening ONLY with these 2 recent beta drivers (13.15 and 13.20)
                Hello, this is exactly what I wrote already at this point [below] - maybe briefly, but these drivers probably have a leak, because for example a running process = game L4D2 will take the maximum available memory and 100% of the time one core CPU and will fill swap - as I wrote below. Unfortunately, the only good drivers are 13.3 - beta3 from 13.4 You will see "miracles" ;P.

                Comment


                • #48
                  Originally posted by ext73 View Post
                  Hello, this is exactly what I wrote already at this point [below] - maybe briefly, but these drivers probably have a leak, because for example a running process = game L4D2 will take the maximum available memory and 100% of the time one core CPU and will fill swap - as I wrote below. Unfortunately, the only good drivers are 13.3 - beta3 from 13.4 You will see "miracles" ;P.

                  http://phoronix.com/forums/showthrea...076#post348076
                  For my 7560D, all drivers were good since 12.9 or 13.1 (the first i installed on it) up to and including 13.6. The 13.15 and 13.20 is unusable.

                  The memory leak doesnt come from the games themselves - they are among the "victims" of the swapping process that begins exactly because large amounts of memory are filled and/or locked down by the driver for some reason. Note that even after killing the processes, that memory is still locked.

                  Also, this is NOT a valve-specific issue because for example Xbmc (which is a media player that uses opengl for its UI and xvba for decoding, has nothing to do with valve) exhibits pretty much the same symptoms, only starting later.

                  My question is - what hardware, kernel version, architecture suffers from this? I have A8-5500 APU, kernel 3.10.4, 64-bit.
                  Additionally, i had to apply that patch to enable all users access to the /proc/ati/ and subfolders, maybe that has something to do with it?

                  Comment


                  • #49
                    I have memory leak also with L4D2 and HL2 but not in DOTA2

                    Distro: arch linux x86_64
                    Kernel: 3.10.3-1
                    CPU: Intel Core i7 2600k
                    GPU: ATI HD6950

                    Comment


                    • #50
                      Originally posted by ext73 View Post
                      Thank you. Thus, below the text already in the patch file. I just saw that there is a problem with permissions [granted only for root], but I did not know where to look . After applying the patch the drivers is 'working'. That's it "works" but you can not say that "working properly". Compared to my opinion the best out 13.3 beta3, which operate directly genius - games such as L4D2, CS: S, Portal, HL2: E2, Dota2, Oil Rush, Xonotic working under this drivers brilliantly! ... and here, as compared to their 13.8 beta1 - X370 tested on AMD APU E-350:

                      - Are slower than 12 to 30% - 12% e.g. at Xonotic, additionally there is not complete liquidity like on 13.3 beta3 - every moment "tears". For example, games in wine 1.6 e.g. LFS operate up to 30% slower than the 13.3 beta 3

                      - Games like L4D2, CS: S, Portal start at some point, "choke" and I saw there some weird dumps [you can not "kill" a game] - congesting the ram memory to the maximum and begins to fill the swap.

                      - Conclusion: the Catalyst 13.4, 13.6 beta [there was input lag - mouse], the version 13.15 [OGL 4.3] where there is no lag but it is the loss of productivity, and in 13.8 beta1 ... These drivers are unusable. Leaving only 13.3 beta3 = maximum stable, efficient, and providing a brilliant system responsiveness [with appropriately optimized kernel]. For example, in my line of Brazos kernels [optimized to run on Athlon II up]:


                      fglrx-13.8-root-fix.patch




                      L4D2 settings on AMD APU E-350:

                      Anizo: 16x
                      VSync: off
                      Shader Details: Medium
                      Details of the effects: medium
                      Details of model/texture: high
                      multi-core rendering: Enabled [but better for the poor E-350 disable this]
                      In fact, give chmod for proc / ati helps temporarily. I tried to rebuild packages firegl_public.c editing the file, but it did not help, maybe I did it wrong. Could someone send me the converted file firegl_public.c?

                      I also tried to add to the spec patch that posted ext73 but it did not work.

                      Comment

                      Working...
                      X