All of lore.kernel.org
 help / color / mirror / Atom feed
From: Chris Wilson <chris@chris-wilson.co.uk>
To: Ben Widawsky <ben@bwidawsk.net>
Cc: intel-gfx@lists.freedesktop.org
Subject: Re: [PATCH 03/18] drm/i915: Fix location of stolen memory register for SandyBridge+
Date: Fri, 02 Nov 2012 08:54:39 +0000	[thread overview]
Message-ID: <eeac1e$4s3j2r@AZSMGA002.ch.intel.com> (raw)
In-Reply-To: <20121101170820.20cb56a2@bwidawsk.net>

On Thu, 1 Nov 2012 17:08:20 -0700, Ben Widawsky <ben@bwidawsk.net> wrote:
> On Sun, 28 Oct 2012 09:48:35 +0000
> Chris Wilson <chris@chris-wilson.co.uk> wrote:
> 
> > On Fri, 26 Oct 2012 14:58:45 -0700, Ben Widawsky <ben@bwidawsk.net> wrote:
> > > On Fri, 19 Oct 2012 18:03:09 +0100
> > > Chris Wilson <chris@chris-wilson.co.uk> wrote:
> > > 
> > > > A few of the earlier registers where enlarged and so the Base Data of
> > > > Stolen Memory Register (BDSM) was pushed to 0xb0.
> > > > 
> > > > Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
> > > > Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
> > > 
> > > This patch seems irrelevant to me. I have a i915_stolen_to_phys which
> > > already looks correct (git blame shows you last updated it in April).
> > > 
> > > Can you help unconfuse me?
> > 
> > As the patch suggests the current registers being used by stolen-to-phys
> > are incorrect for SNB+.
> > -Chris
> > 
> 
> It looks like all you've done here is fix up the bug you left from
> patch2.
> 
> -- 
> Ben Widawsky, Intel Open Source Technology Center
From: Chris Wilson <chris@chris-wilson.co.uk>
Subject: Re: [Intel-gfx] [PATCH 02/18] drm/i915: Fix detection of stolen base for gen2
To: Ben Widawsky <ben@bwidawsk.net>
Cc: intel-gfx@lists.freedesktop.org
In-Reply-To: <20121101165102.3cc9773c@bwidawsk.net>
References: <1350666204-8101-1-git-send-email-chris@chris-wilson.co.uk> <1350666204-8101-2-git-send-email-chris@chris-wilson.co.uk> <20121101165102.3cc9773c@bwidawsk.net>

On Thu, 1 Nov 2012 16:51:02 -0700, Ben Widawsky <ben@bwidawsk.net> wrote:
> On Fri, 19 Oct 2012 18:03:08 +0100
> Chris Wilson <chris@chris-wilson.co.uk> wrote:
> 
> > It was not until the G33 refresh, that a PCI config register was
> > introduced that explicitly said where the stolen memory was. Prior to
> > 865G there was not even a register that said where the end of usable
> > low memory was and where the stolen memory began (or ended depending
> > upon chipset). Before then, one has to look at the BIOS memory maps to
> > find the Top of Memory. Alas that is not exported by arch/x86 and so we
> > have to resort to disabling stolen memory on gen2 for the time being.
> > 
> > Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
> > ---
> >  drivers/gpu/drm/i915/i915_drv.h        |    1 +
> >  drivers/gpu/drm/i915/i915_gem_stolen.c |   69 ++++++++++++++------------------
> >  2 files changed, 31 insertions(+), 39 deletions(-)
> > 
> > diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
> > index 4728d30..687f379 100644
> > --- a/drivers/gpu/drm/i915/i915_drv.h
> > +++ b/drivers/gpu/drm/i915/i915_drv.h
> > @@ -705,6 +705,7 @@ typedef struct drm_i915_private {
> >  		unsigned long gtt_start;
> >  		unsigned long gtt_mappable_end;
> >  		unsigned long gtt_end;
> > +		unsigned long stolen_base; /* limited to low memory (32-bit) */
> >  
> >  		struct io_mapping *gtt_mapping;
> >  		phys_addr_t gtt_base_addr;
> > diff --git a/drivers/gpu/drm/i915/i915_gem_stolen.c b/drivers/gpu/drm/i915/i915_gem_stolen.c
> > index ada2e90..a01ff74 100644
> > --- a/drivers/gpu/drm/i915/i915_gem_stolen.c
> > +++ b/drivers/gpu/drm/i915/i915_gem_stolen.c
> > @@ -43,56 +43,43 @@
> >   * for is a boon.
> >   */
> >  
> > -#define PTE_ADDRESS_MASK		0xfffff000
> > -#define PTE_ADDRESS_MASK_HIGH		0x000000f0 /* i915+ */
> > -#define PTE_MAPPING_TYPE_UNCACHED	(0 << 1)
> > -#define PTE_MAPPING_TYPE_DCACHE		(1 << 1) /* i830 only */
> > -#define PTE_MAPPING_TYPE_CACHED		(3 << 1)
> > -#define PTE_MAPPING_TYPE_MASK		(3 << 1)
> > -#define PTE_VALID			(1 << 0)
> > -
> > -/**
> > - * i915_stolen_to_phys - take an offset into stolen memory and turn it into
> > - *                       a physical one
> > - * @dev: drm device
> > - * @offset: address to translate
> > - *
> > - * Some chip functions require allocations from stolen space and need the
> > - * physical address of the memory in question.
> > - */
> > -static unsigned long i915_stolen_to_phys(struct drm_device *dev, u32 offset)
> > +static unsigned long i915_stolen_to_physical(struct drm_device *dev)
> >  {
> >  	struct drm_i915_private *dev_priv = dev->dev_private;
> >  	struct pci_dev *pdev = dev_priv->bridge_dev;
> >  	u32 base;
> >  
> > -#if 0
> >  	/* On the machines I have tested the Graphics Base of Stolen Memory
> > -	 * is unreliable, so compute the base by subtracting the stolen memory
> > -	 * from the Top of Low Usable DRAM which is where the BIOS places
> > -	 * the graphics stolen memory.
> > +	 * is unreliable, so on those compute the base by subtracting the
> > +	 * stolen memory from the Top of Low Usable DRAM which is where the
> > +	 * BIOS places the graphics stolen memory.
> > +	 *
> > +	 * On gen2, the layout is slightly different with the Graphics Segment
> > +	 * immediately following Top of Memory (or Top of Usable DRAM). Note
> > +	 * it appears that TOUD is only reported by 865g, so we just use the
> > +	 * top of memory as determined by the e820 probe.
> > +	 *
> > +	 * XXX gen2 requires an unavailable symbol and 945gm fails with
> > +	 * its value of TOLUD.
> >  	 */
> > +	base = 0;
> >  	if (INTEL_INFO(dev)->gen > 3 || IS_G33(dev)) {
> > -		/* top 32bits are reserved = 0 */
> > +		/* Read Graphics Base of Stolen Memory directly */
> >  		pci_read_config_dword(pdev, 0xA4, &base);
> > -	} else {
> > -		/* XXX presume 8xx is the same as i915 */
> > -		pci_bus_read_config_dword(pdev->bus, 2, 0x5C, &base);
> > -	}
> > -#else
> > -	if (INTEL_INFO(dev)->gen > 3 || IS_G33(dev)) {
> > -		u16 val;
> > -		pci_read_config_word(pdev, 0xb0, &val);
> > -		base = val >> 4 << 20;
> > -	} else {
> > +#if 0
> > +	} else if (IS_GEN3(dev)) {
> >  		u8 val;
> > +		/* Stolen is immediately below Top of Low Usable DRAM */
> >  		pci_read_config_byte(pdev, 0x9c, &val);
> >  		base = val >> 3 << 27;
> > -	}
> > -	base -= dev_priv->mm.gtt->stolen_size;
> > +		base -= dev_priv->mm.gtt->stolen_size;
> > +	} else {
> > +		/* Stolen is immediately above Top of Memory */
> > +		base = max_low_pfn_mapped << PAGE_SHIFT;
> >  #endif
> > +	}
> >  
> > -	return base + offset;
> > +	return base;
> >  }
> >  
> 
> Comments on this below.
> 
> >  static void i915_warn_stolen(struct drm_device *dev)
> > @@ -117,7 +104,7 @@ static void i915_setup_compression(struct drm_device *dev, int size)
> >  	if (!compressed_fb)
> >  		goto err;
> >  
> > -	cfb_base = i915_stolen_to_phys(dev, compressed_fb->start);
> > +	cfb_base = dev_priv->mm.stolen_base + compressed_fb->start;
> >  	if (!cfb_base)
> >  		goto err_fb;
> >  
> > @@ -130,7 +117,7 @@ static void i915_setup_compression(struct drm_device *dev, int size)
> >  		if (!compressed_llb)
> >  			goto err_fb;
> >  
> > -		ll_base = i915_stolen_to_phys(dev, compressed_llb->start);
> > +		ll_base = dev_priv->mm.stolen_base + compressed_llb->start;
> >  		if (!ll_base)
> >  			goto err_llb;
> >  	}
> > @@ -149,7 +136,7 @@ static void i915_setup_compression(struct drm_device *dev, int size)
> >  	}
> >  
> >  	DRM_DEBUG_KMS("FBC base 0x%08lx, ll base 0x%08lx, size %dM\n",
> > -		      cfb_base, ll_base, size >> 20);
> > +		      (long)cfb_base, (long)ll_base, size >> 20);
> >  	return;
> >  
> >  err_llb:
> > @@ -181,6 +168,10 @@ int i915_gem_init_stolen(struct drm_device *dev)
> >  	struct drm_i915_private *dev_priv = dev->dev_private;
> >  	unsigned long prealloc_size = dev_priv->mm.gtt->stolen_size;
> >  
> > +	dev_priv->mm.stolen_base = i915_stolen_to_physical(dev);
> > +	if (dev_priv->mm.stolen_base == 0)
> > +		return 0;
> > +
> >  	/* Basic memrange allocator for stolen space */
> >  	drm_mm_init(&dev_priv->mm.stolen, 0, prealloc_size);
> >  
> 
> Since I found viewing the diff difficult for this patch, I am going to
> write the before and after (with the #if 0 blocks removed):
> 
> Before:
>         if (INTEL_INFO(dev)->gen > 3 || IS_G33(dev)) {
>                 u16 val;
>                 pci_read_config_word(pdev, 0xb0, &val);
>                 base = val >> 4 << 20;
>         } else {
>                 u8 val;
>                 pci_read_config_byte(pdev, 0x9c, &val);
>                 base = val >> 3 << 27;
>         }
>         base -= dev_priv->mm.gtt->stolen_size;
> 
> After:
>         base = 0;
>         if (INTEL_INFO(dev)->gen > 3 || IS_G33(dev)) {
>                 /* Read Graphics Base of Stolen Memory directly */
>                 pci_read_config_dword(pdev, 0xA4, &base);
> 
> 
> I like that this is a simple helper now to calculate GSM Base. However,
> I am completely baffled by this patch. The read of 0xA4 did exist
> before, but was #if 0'd out appears to be the PCI capabilities pointer
> as of GEN6 (I won't check before that). The before code on the other
> hand at least looks correct for gen6 (BDSM - GGMS).
> 
> This seems like it will definitely break SNB+, which would be bad for
> bisection.

The code as it stands is broken on SNB (it only has a semblance of
working but the address it generates is random and points into active
system memory), so it seems immaterial as to whether it then remained
broken after this patch, with the benefit of then having individual
patches to address the later generations. To allay your fears, this can
be split into a preparatory patch to allow the routine to fail, and then
fail until each is fixed.
-Chris

-- 
Chris Wilson, Intel Open Source Technology Centre

  reply	other threads:[~2012-11-02  8:55 UTC|newest]

Thread overview: 51+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2012-10-19 17:03 [PATCH 01/18] drm: Introduce drm_mm_create_block() Chris Wilson
2012-10-19 17:03 ` [PATCH 02/18] drm/i915: Fix detection of stolen base for gen2 Chris Wilson
2012-11-01 23:51   ` Ben Widawsky
2012-10-19 17:03 ` [PATCH 03/18] drm/i915: Fix location of stolen memory register for SandyBridge+ Chris Wilson
2012-10-26 21:58   ` Ben Widawsky
2012-10-28  9:48     ` Chris Wilson
2012-10-28 17:52       ` Ben Widawsky
2012-11-02  0:08       ` Ben Widawsky
2012-11-02  8:54         ` Chris Wilson [this message]
2012-11-05 13:53           ` Ben Widawsky
2012-10-19 17:03 ` [PATCH 04/18] drm/i915: Avoid clearing preallocated regions from the GTT Chris Wilson
2012-10-26 22:22   ` Ben Widawsky
2012-10-19 17:03 ` [PATCH 05/18] drm: Introduce an iterator over holes in the drm_mm range manager Chris Wilson
2012-11-05 13:41   ` Ben Widawsky
2012-11-05 15:13     ` Chris Wilson
2012-10-19 17:03 ` [PATCH 06/18] drm/i915: Delay allocation of stolen space for FBC Chris Wilson
2012-11-05 13:44   ` Ben Widawsky
2012-11-05 15:24     ` Chris Wilson
2012-10-19 17:03 ` [PATCH 07/18] drm/i915: Defer allocation of stolen memory for FBC until actual first use Chris Wilson
2012-11-05 15:00   ` Ben Widawsky
2012-11-05 15:28     ` Chris Wilson
2012-11-05 15:32       ` Ben Widawsky
2012-10-19 17:03 ` [PATCH 08/18] drm/i915: Allow objects to be created with no backing pages, but stolen space Chris Wilson
2012-10-19 17:03 ` [PATCH 09/18] drm/i915: Differentiate between prime and stolen objects Chris Wilson
2012-10-19 17:03 ` [PATCH 10/18] drm/i915: Support readback of stolen objects upon error Chris Wilson
2012-11-05 15:41   ` Ben Widawsky
2012-10-19 17:03 ` [PATCH 11/18] drm/i915: Handle stolen objects in pwrite Chris Wilson
2012-10-19 17:03 ` [PATCH 12/18] drm/i915: Handle stolen objects for pread Chris Wilson
2012-10-19 17:03 ` [PATCH 13/18] drm/i915: Introduce i915_gem_object_create_stolen() Chris Wilson
2012-11-05 16:32   ` Ben Widawsky
2012-11-05 16:59     ` Chris Wilson
2012-11-05 17:34       ` Ben Widawsky
2012-10-19 17:03 ` [PATCH 14/18] drm/i915: Allocate fbcon from stolen memory Chris Wilson
2012-10-19 17:03 ` [PATCH 15/18] drm/i915: Allocate ringbuffers " Chris Wilson
2012-10-19 17:03 ` [PATCH 16/18] drm/i915: Allocate overlay registers " Chris Wilson
2012-11-05 17:39   ` Ben Widawsky
2012-10-19 17:03 ` [PATCH 17/18] drm/i915: Use a slab for object allocation Chris Wilson
2012-10-24 20:21   ` Paulo Zanoni
2012-11-05 17:49   ` Ben Widawsky
2012-11-05 20:57     ` Chris Wilson
2012-11-07 13:59       ` Ben Widawsky
2012-11-05 18:07   ` Ben Widawsky
2012-11-05 18:10     ` Ben Widawsky
2012-10-19 17:03 ` [PATCH 18/18] drm/i915: Introduce mapping of user pages into video memory (userptr) ioctl Chris Wilson
2012-10-22 22:21   ` Jesse Barnes
2012-10-23  7:20     ` Daniel Vetter
2012-10-23 17:50       ` Eric Anholt
2012-10-26 21:47 ` [PATCH 01/18] drm: Introduce drm_mm_create_block() Ben Widawsky
2012-10-28  9:57   ` Chris Wilson
2012-10-28 18:12     ` Ben Widawsky
2012-10-28 18:14       ` Ben Widawsky

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to='eeac1e$4s3j2r@AZSMGA002.ch.intel.com' \
    --to=chris@chris-wilson.co.uk \
    --cc=ben@bwidawsk.net \
    --cc=intel-gfx@lists.freedesktop.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.