|
Hi Oded!
Thank you for sharing this information!
Tbh, I was surprised that the kbuild test robot did cc Andrew and others, as it's my private repo, and I didn't send these patches publicly yet. Anyway, thank you for looking into the problem.
Thanks! Roman From: Oded Gabbay <oded.gabbay@xxxxxxxxx>
Sent: Saturday, March 2, 2019 1:09:09 AM To: Roman Gushchin; Andrew Morton Cc: Linux Memory Management List; Johannes Weiner Subject: Re: [rgushchin:release_percpu 321/401] ERROR: "dma_fence_release" [drivers/misc/habanalabs/habanalabs.ko] undefined! On Sat, Mar 2, 2019 at 3:42 AM kbuild test robot <lkp@xxxxxxxxx> wrote:
> > Hi Andrew, > > First bad commit (maybe != root cause): > > tree: https://github.com/rgushchin/linux.git release_percpu > head: 8b287c57af99a4642cdf70b3b1a5ab1d90877bba > commit: 773ae09cc9c61b8f7dc983e9a1d4ce5abbd6339d [321/401] linux-next-rejects > config: x86_64-randconfig-m3-02280836 (attached as .config) > compiler: gcc-7 (Debian 7.4.0-5) 7.4.0 > reproduce: > git checkout 773ae09cc9c61b8f7dc983e9a1d4ce5abbd6339d > # save the attached .config to linux build tree > make ARCH=x86_64 > > All errors (new ones prefixed by >>): > > >> ERROR: "dma_fence_release" [drivers/misc/habanalabs/habanalabs.ko] undefined! > >> ERROR: "dma_fence_init" [drivers/misc/habanalabs/habanalabs.ko] undefined! > >> ERROR: "dma_fence_signal" [drivers/misc/habanalabs/habanalabs.ko] undefined! > >> ERROR: "dma_fence_default_wait" [drivers/misc/habanalabs/habanalabs.ko] undefined! > >> ERROR: "dma_fence_wait_timeout" [drivers/misc/habanalabs/habanalabs.ko] undefined! > > --- > 0-DAY kernel test infrastructure Open Source Technology Center > https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.01.org_pipermail_kbuild-2Dall&d=DwIBaQ&c=5VD0RTtNlTh3ycd41b3MUw&r=jJYgtDM7QT-W-Fz_d29HYQ&m=HPA84jVOmuSotww9zAhO59SftxRFVUYNTjerZypaw8k&s=s7A8kZ6RjzQwYxsjp9zi0KTzwWbmVFkEPUDEvtz7rzA&e= Intel Corporation Hi Roman and Andrew. That error comes from the original patch-set of the driver, where a there was missing "select DMA_SHARED_BUFFER" in Kconfig. So for certain configurations, the build would fail. This is already fixed in the current char-misc-next tree of gkh. Thanks, Oded |
