You are not logged in.
Pages: 1
hi,
I've written a program for a school presentation which simulates a magnetic pendulum. I use OpenGL for calculating and rendering the basins. To keep it short: I can calculate with different precisions. The more precise the calculation should be, the more time the shader program needs to finish. But when the shader program needs to much time my radeon driver crashes
here is the dmesg output
[ 193.001928] WARNING: CPU: 1 PID: 512 at drivers/gpu/drm/radeon/radeon_object.c:84 radeon_ttm_bo_destroy+0xf1/0x100 [radeon]()
[ 193.001930] Modules linked in: ctr ccm arc4 ath9k ath9k_common snd_hda_codec_realtek snd_hda_codec_generic snd_hda_codec_hdmi amdkfd amd_iommu_v2 ath9k_hw
ath snd_hda_intel snd_hda_controller led_class amd_freq_sensitivity kvm_amd kvm crct10dif_pclmul crc32_pclmul crc32c_intel ghash_clmulni_intel mac80211 aesni_
intel radeon ttm drm_kms_helper cfg80211 aes_x86_64 snd_hda_codec lrw gf128mul drm rfkill joydev r8169 mii snd_hwdep snd_pcm glue_helper ablk_helper cryptd ev
dev mousedev mac_hid serio_raw pcspkr snd_timer k10temp fam15h_power shpchp hwmon snd edac_core i2c_piix4 i2c_algo_bit i2c_core edac_mce_amd soundcore tpm_tis
tpm button acpi_cpufreq processor sch_fq_codel ext4 crc16 mbcache jbd2 hid_generic usbhid hid sd_mod atkbd libps2 ahci libahci xhci_pci ohci_pci ohci_hcd ehc
i_pci xhci_hcd ehci_hcd
[ 193.001980] libata usbcore usb_common scsi_mod i8042 serio
[ 193.001987] CPU: 1 PID: 512 Comm: pendel Tainted: G W 4.0.2-1-ARCH #1
[ 193.001989] Hardware name: To Be Filled By O.E.M. To Be Filled By O.E.M./AM1H-ITX, BIOS P1.00 02/19/2014
[ 193.001991] 0000000000000000 0000000007e90f0d ffff88008ecc3b58 ffffffff81571be3
[ 193.001994] 0000000000000000 0000000000000000 ffff88008ecc3b98 ffffffff81074dda
[ 193.001997] 000000000000000e ffff88008eefec68 ffff88008eefec00 ffffffffffffffff
[ 193.002001] Call Trace:
[ 193.002005] [<ffffffff81571be3>] dump_stack+0x4c/0x6e
[ 193.002009] [<ffffffff81074dda>] warn_slowpath_common+0x8a/0xc0
[ 193.002013] [<ffffffff81074f0a>] warn_slowpath_null+0x1a/0x20
[ 193.002027] [<ffffffffa0481eb1>] radeon_ttm_bo_destroy+0xf1/0x100 [radeon]
[ 193.002032] [<ffffffffa040c946>] ttm_bo_release_list+0xa6/0x190 [ttm]
[ 193.002039] [<ffffffffa040d7f8>] ttm_bo_release+0x208/0x2e0 [ttm]
[ 193.002044] [<ffffffffa040d8f9>] ttm_bo_unref+0x29/0x30 [ttm]
[ 193.002058] [<ffffffffa0482419>] radeon_bo_unref+0x39/0x70 [radeon]
[ 193.002074] [<ffffffffa049777b>] radeon_gem_object_free+0x4b/0x70 [radeon]
[ 193.002081] [<ffffffffa0318c07>] drm_gem_object_free+0x27/0x40 [drm]
[ 193.002090] [<ffffffffa0318d40>] drm_gem_object_handle_unreference_unlocked+0x120/0x130 [drm]
[ 193.002097] [<ffffffffa0318ec4>] drm_gem_handle_delete+0xb4/0x100 [drm]
[ 193.002105] [<ffffffffa0319275>] drm_gem_close_ioctl+0x25/0x30 [drm]
[ 193.002112] [<ffffffffa0319caf>] drm_ioctl+0x1df/0x680 [drm]
[ 193.002124] [<ffffffffa046104c>] radeon_drm_ioctl+0x4c/0x80 [radeon]
[ 193.002128] [<ffffffff811ebb76>] do_vfs_ioctl+0x2c6/0x4d0
[ 193.002134] [<ffffffff811da1fe>] ? ____fput+0xe/0x10
[ 193.002137] [<ffffffff811ebe01>] SyS_ioctl+0x81/0xa0
[ 193.002141] [<ffffffff81577509>] system_call_fastpath+0x12/0x17
[ 193.002143] ---[ end trace eb88cfdb5a56a3c8 ]---when I restart Xorg after the crash the radeon driver is replaced by the vmware driver
hostname% glxinfo | grep -i opengl
OpenGL vendor string: VMware, Inc.
OpenGL renderer string: Gallium 0.4 on llvmpipe (LLVM 3.6, 128 bits)
OpenGL version string: 3.0 Mesa 10.5.5
OpenGL shading language version string: 1.30
OpenGL context flags: (none)
OpenGL extensions:Last edited by rejak (2015-05-15 22:28:06)
Offline
Though I am of NO HELP on this issue, I would think posting your code might help troubleshoot things. There are experts here that could at least see if it's an obvious coding problem or maybe even test it on their own machines.
Matt
"It is very difficult to educate the educated."
Offline
I don't think you understand what a "tainted kernel" is. Try looking it up with google.
Offline
I had to reread your original post a few times... ![]()
when the shader program needs to much time Xorg crashes.
THIS is the real problem, right? Why is your software causing Xorg to crash? Is it eating up all your memory or something?
Offline
I had to reread your original post a few times...
rejak wrote:when the shader program needs to much time Xorg crashes.
THIS is the real problem, right? Why is your software causing Xorg to crash? Is it eating up all your memory or something?
My program is causing the radeon module to crash, and the radeon module causes Xorg to crash. But a userspace program shouldn't be able to crash a kernel module.
Offline
I don't think you understand what a "tainted kernel" is. Try looking it up with google.
thx, I've corrected the thread
Offline
It might be a bug in the driver and it should be reported upstream. Given that you have a way to reliably trigger the problem, your magnetic pendulum simulation code, it should be relatively easy to get someone to look at the problem assuming you can share your code.
R00KIE
Tm90aGluZyB0byBzZWUgaGVyZSwgbW92ZSBhbG9uZy4K
Offline
Pages: 1