On 06.05.2025 10:31, Roger Pau Monne wrote:
> The current underlying implementation of GNTTABOP_cache_flush on x86 won't
> work as expected. The provided {clean,invalidate}_dcache_va_range()
> helpers only do a local pCPU cache flush, so the cache of previous pCPUs
> where the vCPU might have run are not flushed.
>
> However instead of attempting to fix this, make the GNTTABOP_cache_flush
> operation only available to ARM. There are no known users on x86, the
> implementation is broken, and other architectures don't have grant-table
> support yet.
>
> With that operation not implemented on x86, the related
> {clean,invalidate}_dcache_va_range() helpers can also be removed.
>
> Fixes: f62dc81c2df7 ("x86: introduce more cache maintenance operations")
> Fixes: 18e8d22fe750 ("introduce GNTTABOP_cache_flush")
> Signed-off-by: Roger Pau Monné <roger....@citrix.com>
Ah, here we go. I think this is what we want, without patch 3. It will want
to come with a CHANGELOG entry, though.
> --- a/xen/common/grant_table.c
> +++ b/xen/common/grant_table.c
> @@ -940,6 +940,7 @@ static void reduce_status_for_pin(struct domain *rd,
> gnttab_clear_flags(rd, clear_flags, status);
> }
>
> +#ifdef CONFIG_ARM
Better introduce a new Kconfig setting (e.g. HAS_GRANT_CACHE_FLUSH) right
away, in case RISC-V and/or PPC would also want such behavior?
Jan
