Re: [Qemu-devel] [PULL 0/8] Linux user for 2.12 patches

[Peter Maydell]

On 19 February 2018 at 17:10, Laurent Vivier <address@hidden> wrote:
> The following changes since commit f0fa81767555fe2c4b5f8c9e0725a80eac1d7f56:
>
>   cuda.h: Fix multiple typedef (2018-02-16 18:38:21 +0000)
>
> are available in the Git repository at:
>
>   git://github.com/vivier/qemu.git tags/linux-user-for-2.12-pull-request
>
> for you to fetch changes up to f352a021329f73d0219018cdfb39369687e8cadd:
>
>   linux-user: MIPS set cpu to r6 CPU if binary is R6 (2018-02-18 18:52:32 
> +0100)
>
> ----------------------------------------------------------------
>
> ----------------------------------------------------------------

Hi. This breaks one of my linux-user smoke tests:

/home/petmay01/linaro/qemu-for-merges/build/all-linux-static/sparc32plus-linux-user/qemu-sparc32plus
-L ./gnemul/qemu-sparc sparc/ls -l dummyfile
unable to find CPU model 'Fujitsu MB86904'

This is because the include path for linux-user picks the
guest-architecture subdirectory for the TARGET_ABI_DIR,
which for sparc32plus is "sparc", but the set of CPUs compiled
in depends on the TARGET_ARCH, which is "sparc64", and
sparc64 doesn't include that 32-bit Fujitsu CPU.
The linux-user/sparc/target_elf.h needs to have code to handle
both the pure-32-bit and 32-bit-ABI-on-64-bit-CPU cases,
the same way that linux-user/ppc/target_elf.h does.

thanks
-- PMM