Re: [PATCH] arm64: fix build regression in of_path_of_nvme

[Eric Snowberg]

> On Feb 27, 2018, at 5:06 AM, Daniel Kiper <address@hidden> wrote:
> 
> On Mon, Feb 26, 2018 at 10:59:24AM +0100, Joakim Bech wrote:
>> On Mon, Feb 26, 2018 at 10:45:46AM +0100, Daniel Kiper wrote:
>>> On Mon, Feb 26, 2018 at 09:57:34AM +0100, Joakim Bech wrote:
>>>> The of_path_of_nvme function introduced a build regression:
>>>>    grub-core/osdep/linux/ofpath.c:365:21: error: comparison between pointer
>>>>    and zero character constant [-Werror=pointer-compare]
>>>>       if ((digit_string != '\0') && (*part_end == 'p'))
>>>> 
>>>> Update digit_string to compare against the char instead of the pointer.
>>>> 
>>>> Signed-off-by: Joakim Bech <address@hidden>
>>> 
>>> CC-ing Vladimir.
>>> 
>>> Pushed with some commit message changes. Thanks! However, this raises
>>> the question: why this code is build on ARM? Should we do not do that?
>>> 
>> My bad here Daniel, I was building for the HiKey device which uses the
>> AArch64 GCC for most of its software components, but you are right, when
>> checking closer to the output from our builds I can see that we are
>> using the regular x86 toolchain when compiling grub. So a more correct
> 
> 8-) Well... Eric, should not we build this code on SPARC only? I do not
> think it makes much sense to have that thing on platforms not supporting
> Open Firmware…

Some of the ofpath code is common across all Open Firmware platforms.  There 
are some unique differences in naming that have come about over time. I don’t 
know how PPC or x86 do their NVMe Open Firmware path naming.  I left it this 
way since once those architectures added NVMe support, hopefully most of the 
code could be reused. This is similar to how SAS originally only worked on PPC 
and didn’t work on SPARC. If you want to make it SPARC only at the moment, I’m 
ok with that.