Red Hat Bugzilla – Bug 520186
kernel gets DMA related warnings in e100 driver
Last modified: 2009-12-04 13:30:48 EST
Created attachment 359110 [details]
dmesg output from 2.6.31-0.185.rc7.git6.fc12.x86_64
Description of problem:
Kernels from 2.6.31 series register warnings as follows:
e100 0000:00:0e.0: DMA-API: device driver syncs DMA memory with different direction [device address=0x0000000018c320b2] [size=1534 bytes] [mapped with DMA_BIDIRECTIONAL] [synced with DMA_FROM_DEVICE]
That followed by a call trace from an IRQ handler. It does not appear to be harmful although an ethernet traffic here is not too excessive.
Version-Release number of selected component (if applicable):
2.6.31-0.185.rc7.git6.fc12.x86_64 and earlier
On every boot
dmesg output attached
So far that complaint is consistent across all 2.6.31-... kernels up and including 2.6.31-0.204.rc9.fc12.x86_64
This does not stop. The current kernel 2.6.31-40.fc12.x86_64, and all versions in between, are getting the same warnings and traces on every boot. Yes, this is from lib/dma-debug.c:917 check_sync+0x252/0x449() and a trace is followed by:
[<ffffffffa017bb3d>] pci_map_single+0xd3/0xfa [e100]
[<ffffffffa017e707>] e100_rx_alloc_skb+0x91/0x14a [e100]
[<ffffffffa017e872>] e100_rx_alloc_list+0xb2/0x138 [e100]
[<ffffffffa017e921>] e100_up+0x29/0x110 [e100]
Seeing on i686 as well:
Oct 12 14:24:08 localhost kernel: WARNING: at lib/dma-debug.c:917 check_sync+0x23c/0x413() (Not tainted)
Oct 12 14:24:08 localhost kernel: Hardware name: Dimension 4700
Oct 12 14:24:08 localhost kernel: e100 0000:03:08.0: DMA-API: device driver syncs DMA memory with different direction [device address=0x0000000033318852] [size=1534 bytes] [mapped with DMA_BIDIRECTIONAL] [synced with DMA_FROM_DEVICE]
Oct 12 14:24:08 localhost kernel: Modules linked in: autofs4 ipv6 p4_clockmod dm_multipath uvcvideo videodev v4l1_compat snd_usb_audio snd_usb_lib snd_rawmidi snd_seq_device snd_hwdep dcdbas ppdev e100 mii snd_intel8x0 parport_pc snd_ac97_codec ac97_bus parport snd_pcm i2c_i801 snd_timer iTCO_wdt snd iTCO_vendor_support soundcore snd_page_alloc radeon ttm drm_kms_helper drm i2c_algo_bit i2c_core [last unloaded: microcode]
Oct 12 14:24:08 localhost kernel: Pid: 1068, comm: modem-manager Not tainted 126.96.36.199-56.fc12.i686.PAE #1
In this case while technically it's a bug, it can't be fixed without major changes to the driver that would involve using an unsupported operating mode or copying every packet before processing it. The driver has been fixed so it operates correctly even though there is still a warning. See:
This bug appears to have been reported against 'rawhide' during the Fedora 12 development cycle.
Changing version to '12'.
More information and reason for this action is here:
The VERIFIED, FAILS_QA and RELEASE_PENDING bug states are not used by Fedora (they are used in the RHEL process).
From what I can understand, CLOSED UPSTREAM is much better resolution here.
Fedora Bugzappers volunteer triage team