On 11/12/15 23:29, Jonas Gorski wrote: > On Fri, Dec 11, 2015 at 11:24 PM, Simon Arlott <simon@xxxxxxxxxxx> wrote: >> On 11/12/15 22:02, Jonas Gorski wrote: >>> On Fri, Dec 11, 2015 at 10:54 PM, Simon Arlott <simon@xxxxxxxxxxx> wrote: >>>> Broadcom BCM963xx boards have multiple nvram variants across different >>>> SoCs with additional checksum fields added whenever the size of the >>>> nvram was extended. >>>> >>>> Add this structure as a header file so that multiple drivers and userspace >>>> can use it. >>>> >>>> Signed-off-by: Simon Arlott <simon@xxxxxxxxxxx> >>>> --- >>>> v3: Fix includes/type names, add comments explaining the nvram struct. >>>> >>>> v2: Use external struct bcm963xx_nvram definition for bcm963268part. ... >>>> diff --git a/include/uapi/linux/bcm963xx_nvram.h b/include/uapi/linux/bcm963xx_nvram.h >>>> new file mode 100644 >>>> index 0000000..2dcb307 >>>> --- /dev/null >>>> +++ b/include/uapi/linux/bcm963xx_nvram.h >>> >>> Why uapi? The nvram layout isn't really enforced to be that way, and >>> at least Huawei uses a modified one on some devices (in case you >>> wondered why bcm63xx doesn't fail a crc32-"broken" one), so IMHO it >>> should be kept for in-kernel use only. >> >> Because Florian suggested include/uapi/linux/bcm963xx_nvram.h; I could >> move it to include/linux/ instead if this is preferred. ... >>>> + * Do not use the __reserved fields, especially not as an offset for CRC >>>> + * calculations (use BCM963XX_NVRAM_*_SIZE instead). These may be removed or >>>> + * repositioned. > > Because I just saw that: Nobody will read that. ;p I'll move this to include/linux/bcm963xx_nvram.h and omit the linux-api mailing list when I next send the patch series. -- Simon Arlott
