On 2/5/19 10:18 AM, Peter Maydell wrote:
> In commit f7b78602fdc6c6e4be we added the CPU cluster number to the
> cflags field of the TB hash; this included adding it to the value
> kept in tb->cflags, since we pass that field directly into the hash
> calculation in some places. Unfortunately we forgot to check whether
> other parts of the code were doing comparisons against tb->cflags
> that would need to be updated.
>
> It turns out that there is exactly one such place: the
> tb_lookup__cpu_state() function checks whether the TB it has
> found in the tb_jmp_cache has a tb->cflags matching the cf_mask
> that is passed in. The tb->cflags has the cluster_index in it
> but the cf_mask does not.
>
> Hoist the "add cluster index to the cf_mask" code up from
> tb_htable_lookup() to tb_lookup__cpu_state() so it can be considered
> in the "did this TB match in the jmp cache" condition, as well as
> when we do the full hash lookup by physical PC, flags, etc.
> (tb_htable_lookup() is only called from tb_lookup__cpu_state(),
> so this change doesn't require any further knock-on changes.)
>
> Fixes: f7b78602fdc6c6e4be ("accel/tcg: Add cluster number to TCG TB hash")
> Reported-by: Howard Spoelstra <hsp.c...@gmail.com>
> Reported-by: Cleber Rosa <cr...@redhat.com>
> Signed-off-by: Peter Maydell <peter.mayd...@linaro.org>
> ---
> Does anybody know why tb_lookup__cpu_state() has that odd
> double-underscore in the middle of its name?
>
> Since the jmp_cache is per-vcpu we know that we're always going
> to match on the cluster_index, so the other option would be to
> leave the cluster_index bits out of the comparison, and leave the
> "fold in cluster index to cf_mask" code in tb_htable_lookup().
> Or we could require the callers of tb_lookup__cpu_state() to all
> provide the cluster index, but that's more places to change,
> so I prefer this.
I can confirm the performance regression I experienced is fixed by this
patch.
Tested-by: Cleber Rosa <cr...@redhat.com>
