linux-kernel - Re: [RFC PATCH 2/2] drm/ttm: downgrade cached to write

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <d721003aca5b023feee8ab63a4ac29944f5b4549.camel@icenowy.me>
Date: Sun, 30 Jun 2024 14:52:47 +0800
From: Icenowy Zheng <uwu@...nowy.me>
To: Jiaxun Yang <jiaxun.yang@...goat.com>, Christian Koenig
 <christian.koenig@....com>, Huang Rui <ray.huang@....com>, Maarten
 Lankhorst <maarten.lankhorst@...ux.intel.com>, Maxime Ripard
 <mripard@...nel.org>,  Thomas Zimmermann <tzimmermann@...e.de>, David
 Airlie <airlied@...il.com>, Daniel Vetter <daniel@...ll.ch>
Cc: dri-devel@...ts.freedesktop.org, linux-kernel@...r.kernel.org
Subject: Re: [RFC PATCH 2/2] drm/ttm: downgrade cached to write_combined
 when snooping not available

在 2024-06-29星期六的 20:57 +0100，Jiaxun Yang写道：
> 
> 
> 在2024年6月29日六月 上午6:22，Icenowy Zheng写道：
> [...]
> > @@ -302,6 +302,10 @@ pgprot_t ttm_io_prot(struct ttm_buffer_object
> > *bo, 
> > struct ttm_resource *res,
> >                 caching = res->bus.caching;
> >         }
> > 
> > +       /* Downgrade cached mapping for non-snooping devices */
> > +       if (!bo->bdev->dma_coherent && caching == ttm_cached)
> > +               caching = ttm_write_combined;
> Hi Icenowy,
> 
> Thanks for your patch! You saved many non-coh PCIe host
> implementations a day!.
> 
> Unfortunately I don't think we can safely ttm_cached to
> ttm_write_comnined, we've
> had enough drama with write combine behaviour on all different
> platforms.

I think on these platforms, ttm_write_combined should be prevented to
be mapped to pgprot_writecombine instead, but downgrade ttm_cached to
ttm_write_combined (and then being treated same as ttm_uncached) is
acceptable.

> 
> See drm_arch_can_wc_memory in drm_cache.h.
> 
> Thanks 
> 
> > +
> >         return ttm_prot_from_caching(caching, tmp);
> >  }
> >  EXPORT_SYMBOL(ttm_io_prot);
> > diff --git a/drivers/gpu/drm/ttm/ttm_tt.c
> > b/drivers/gpu/drm/ttm/ttm_tt.c
> > index 7b00ddf0ce49f..3335df45fba5e 100644
> > --- a/drivers/gpu/drm/ttm/ttm_tt.c
> > +++ b/drivers/gpu/drm/ttm/ttm_tt.c
> > @@ -152,6 +152,10 @@ static void ttm_tt_init_fields(struct ttm_tt
> > *ttm,
> >                                enum ttm_caching caching,
> >                                unsigned long extra_pages)
> >  {
> > +       /* Downgrade cached mapping for non-snooping devices */
> > +       if (!bo->bdev->dma_coherent && caching == ttm_cached)
> > +               caching = ttm_write_combined;
> > +
> >         ttm->num_pages = (PAGE_ALIGN(bo->base.size) >> PAGE_SHIFT)
> > + extra_pages;
> >         ttm->page_flags = page_flags;
> >         ttm->dma_address = NULL;
> > diff --git a/include/drm/ttm/ttm_caching.h
> > b/include/drm/ttm/ttm_caching.h
> > index a18f43e93abab..f92d7911f50e4 100644
> > --- a/include/drm/ttm/ttm_caching.h
> > +++ b/include/drm/ttm/ttm_caching.h
> > @@ -47,7 +47,8 @@ enum ttm_caching {
> > 
> >         /**
> >          * @ttm_cached: Fully cached like normal system memory,
> > requires that
> > -        * devices snoop the CPU cache on accesses.
> > +        * devices snoop the CPU cache on accesses. Downgraded to
> > +        * ttm_write_combined when the snooping capaiblity is
> > missing.
> >          */
> >         ttm_cached
> >  };
> > -- 
> > 2.45.2
>