Re: [PATCH 1/3] x86/shadow: unlock P2M slightly earlier in shadow_track_dirty_vram()

["Roger Pau Monné" <roger.pau@citrix.com>]

On Tue, Feb 03, 2026 at 05:49:35PM +0100, Jan Beulich wrote:
> There's no need to call vfree() with the lock still held.
> 
> Signed-off-by: Jan Beulich <jbeulich@suse.com>

Acked-by: Roger Pau Monné <roger.pau@citrix.com>

> ---
> In fact for the purposes of the function the P2M lock could really be
> obtained merely in read mode, and it could be dropped immediately in both
> the main "if()" and its "else if()". If only there wasn't the error
> handling after copy_to_guest(): Dropping the paging lock ahead of that
> call, we rely solely on the P2M lock to also guard the changing of
> d->arch.hvm.dirty_vram.sh and what it points to. Question is why dropping
> the paging lock (but continuing to hold the P2M lock) is necessary there
> in the first place.

I wouldn't spend a lot of time trying to optimize this, we already
know HVM shadow is ATM not very optimized, and generally recommend
HAP.

If we could turn the p2m lock into read-mode, maybe at the expense of
expanding the paging locked region that would likely be slightly better?

Thanks, Roger.