From: Peter Huewe <peterhuewe@gmx.de> This patch fixes a build failure for various arm based defconfigs [1][2][3] and maybe other architectures/configs. The build failure was introduced by the sh specific patch [4] "serial: sh-sci: Add DMA support" by Guennadi Liakhovetski Patch against linux-next of 20100309 References: [1] http://kisskb.ellerman.id.au/kisskb/buildresult/2248992/ [2] http://kisskb.ellerman.id.au/kisskb/buildresult/2248996/ [3] http://kisskb.ellerman.id.au/kisskb/buildresult/2248998/ [4] http://git.kernel.org/?p=linux/kernel/git/sfr/linux-next.git;a=commit;h=73a19e4c030190... Signed-off-by: Peter Huewe <peterhuewe@gmx.de> --- include/linux/serial_sci.h | 4 ++++ 1 files changed, 4 insertions(+), 0 deletions(-) diff --git a/include/linux/serial_sci.h b/include/linux/serial_sci.h index 1b177d2..193d4bf 100644 --- a/include/linux/serial_sci.h +++ b/include/linux/serial_sci.h @@ -2,7 +2,9 @@ #define __LINUX_SERIAL_SCI_H #include <linux/serial_core.h> +#ifdef CONFIG_SERIAL_SH_SCI_DMA #include <asm/dmaengine.h> +#endif /* * Generic header for SuperH SCI(F) (used by sh/sh64/h8300 and related parts) @@ -30,8 +32,10 @@ struct plat_sci_port { upf_t flags; /* UPF_* flags */ char *clk; /* clock string */ struct device *dma_dev; +#ifdef CONFIG_SERIAL_SH_SCI_DMA enum sh_dmae_slave_chan_id dma_slave_tx; enum sh_dmae_slave_chan_id dma_slave_rx; +#endif }; #endif /* __LINUX_SERIAL_SCI_H */ -- 1.6.4.4 --
Please Cc linux-sh@vger.kernel.org so these things make it in to Applied, thanks. --
Ok, I understand, this breakage is only in linux-next, and only for shmobile targets, which wasn't there as I was preparing the patch. However, what I did oversee is the H8300 arch, which also uses the sh-sci.c, and they are, probably, also broken in the mainline now. So, can Thanks Guennadi --- Guennadi Liakhovetski, Ph.D. Freelance Open-Source Software Developer http://www.open-technology.de/ --
Yes, it's in my .34 queue. H8300 doesn't actually matter though, since they've never converted in-tree to using sh-sci through the driver model. It likely hasn't been tested on H8 since 2.5, so I'm not exactly losing sleep over it. --
