Re: [PATCH v3 2/3] bsd-user: Propagate alignment argument to mmap_find_vma()

[Richard Henderson <richard.henderson@linaro.org>]

On 3/7/25 05:09, Philippe Mathieu-Daudé wrote:
> Propagate the alignment to mmap_find_vma(), effectively
> embedding mmap_find_vma_aligned() within mmap_find_vma().
> 
> Since we ignore the alignment in do_bsd_shmat(), leave a
> FIXME comment.
> 
> Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
> ---
>   bsd-user/bsd-mem.h |  2 +-
>   bsd-user/qemu.h    |  2 +-
>   bsd-user/mmap.c    | 10 ++--------
>   3 files changed, 4 insertions(+), 10 deletions(-)
> 
> diff --git a/bsd-user/bsd-mem.h b/bsd-user/bsd-mem.h
> index f5ec0de24ca..87219da2919 100644
> --- a/bsd-user/bsd-mem.h
> +++ b/bsd-user/bsd-mem.h
> @@ -372,7 +372,7 @@ static inline abi_long do_bsd_shmat(int shmid, abi_ulong shmaddr, int shmflg)
>           } else {
>               abi_ulong mmap_start;
>   
> -            mmap_start = mmap_find_vma(0, shm_info.shm_segsz);
> +            mmap_start = mmap_find_vma(0, shm_info.shm_segsz, 0 /* FIXME??? */);

It's not really ignoring the alignment, but not requiring alignment above page size.

Traditionally, the alignment for shmat should be SHMLBA.
But in current freebsd sources,

sys/sys/shm.h:#define SHMLBA      PAGE_SIZE /* Segment low boundary address multiple */

there are no crazy broken old architectures to worry about.


r~