linuxppc-dev.lists.ozlabs.org archive mirror
 help / color / mirror / Atom feed
From: Jeff Hartmann <jhartmann@valinux.com>
To: Gareth Hughes <gareth@valinux.com>
Cc: Dan Malek <dan@mvista.com>,
	michdaen@iiic.ethz.ch, dri-devel@lists.sourceforge.net,
	linuxppc-dev@lists.linuxppc.org
Subject: Re: [Dri-devel] Re: PPC Lockup (ati-pcigart-branch)
Date: Fri, 19 Jan 2001 09:48:50 -0700	[thread overview]
Message-ID: <3A686FF2.8000204@valinux.com> (raw)
In-Reply-To: 3A67E467.74544C4C@valinux.com

[-- Attachment #1: Type: text/plain, Size: 1863 bytes --]

Gareth Hughes wrote:

> Dan Malek wrote:
>
>> Michel Dänzer wrote:
>>
>>> [CC'ing linuxppc-dev, hopefully someone there knows what might be up...]
>>>
>>> This is where it dies:
>>>
>>>         /* FIXME: We should really have a kernel call for this...
>>>          */
>>>         entry->virtual = __vmalloc( (pages << PAGE_SHIFT),
>>>                                     GFP_KERNEL,
>>>                                     PAGE_KERNEL);
>>
>> This isn't very much information, but only one thing can really
>> be wrong........
>>
>> What is the value of 'pages'?  I suspect it is huge (and perhaps
>> wrong).  The GFP_KERNEL flag will cause vmalloc() to wait for pages
>> to become available (i.e. it will swap other things out).  If this
>> value in incorrect, this call will wait forever for pages that are
>> never going to arrive.  Worse, it is going to keep sucking up pages
>> and holding them, so nothing else is going to run either.  Is "pages"
>> really the number of pages, or a size that was never converted to pages?
>
>
> Pages is definitely the number of pages (at least when I wrote the
> code...).

It is definitely the number of pages.  Also this function works
correctly on an ia32, I'll post the file where this code is executed in
case there is a simple error which the PPC people would see.

>
>
>> ...and for the 'FIXME' comment, you want a function that does
>> what? vmalloc?  Why don't you just call it (or one of the more
>> appropriate variants if necessary)?
>
>
> This was originally when I was passing a flag to make the memory
> uncached.  This wasn't needed, and we really should be using
> vmalloc_32(...) instead (which will result in exactly the same code, but
> it is cleaner).

I agree, I haven't gotten around to code cleanup on this branch yet though.

-Jeff
(The function in question is drm_sg_alloc starting on line 64.)


[-- Attachment #2: scatter.c --]
[-- Type: text/plain, Size: 5607 bytes --]

/* scatter.c -- IOCTLs to manage scatter/gather memory -*- linux-c -*-
 * Created: Mon Dec 18 23:20:54 2000 by gareth@valinux.com
 *
 * Copyright 2000 VA Linux Systems, Inc., Sunnyvale, California.
 * All Rights Reserved.
 *
 * Permission is hereby granted, free of charge, to any person obtaining a
 * copy of this software and associated documentation files (the "Software"),
 * to deal in the Software without restriction, including without limitation
 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
 * and/or sell copies of the Software, and to permit persons to whom the
 * Software is furnished to do so, subject to the following conditions:
 *
 * The above copyright notice and this permission notice (including the next
 * paragraph) shall be included in all copies or substantial portions of the
 * Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
 * PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR
 * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
 * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
 * DEALINGS IN THE SOFTWARE.
 *
 * Authors:
 *	Gareth Hughes <gareth@valinux.com>
 *
 */

#define __NO_VERSION__
#include <linux/config.h>
#include "drmP.h"
#include <linux/wrapper.h>

#define DEBUG_SCATTER 1

static void drm_sg_cleanup( drm_sg_mem_t *entry )
{
	int i;

	for ( i = 0 ; i < entry->pages ; i++ ) {
		ClearPageReserved( entry->pagelist[i] );
	}

	vfree( entry->virtual );

	drm_free( entry->pagelist,
		  entry->pages * sizeof(*entry->pagelist),
		  DRM_MEM_PAGES );
	drm_free( entry,
		  sizeof(*entry),
		  DRM_MEM_SGLISTS );
}

static inline long usec( void )
{
   struct timeval tv;

   do_gettimeofday( &tv );

   return (tv.tv_sec & 0x7ff) * 1000000 + tv.tv_usec;
}

int drm_sg_alloc( struct inode *inode, struct file *filp,
		  unsigned int cmd, unsigned long arg )
{
	drm_file_t *priv = filp->private_data;
	drm_device_t *dev = priv->dev;
	drm_scatter_gather_t request;
	drm_sg_mem_t *entry;
	unsigned long pages;
	pgd_t *pgd;
	pmd_t *pmd;
	pte_t *pte;
	int i, j;
	DRM_INFO( "%s\n", __FUNCTION__ );

	if ( dev->sg ) return -EINVAL;

	if ( copy_from_user( &request,
			     (drm_scatter_gather_t *)arg,
			     sizeof(request) ) )
		return -EFAULT;

	entry = drm_alloc( sizeof(*entry), DRM_MEM_SGLISTS );
	if ( !entry ) return -ENOMEM;

   	memset( entry, 0, sizeof(*entry) );

	pages = (request.size + PAGE_SIZE - 1) / PAGE_SIZE;
	DRM_DEBUG( "sg size=%ld pages=%ld\n", request.size, pages );

	entry->pages = pages;
	entry->pagelist = drm_alloc( pages * sizeof(*entry->pagelist),
				     DRM_MEM_PAGES );
	if ( !entry->pagelist ) {
		drm_free( entry, sizeof(*entry), DRM_MEM_SGLISTS );
		return -ENOMEM;
	}

	/* FIXME: We should really have a kernel call for this...
	 */
	entry->virtual = __vmalloc( (pages << PAGE_SHIFT),
				    GFP_KERNEL,
				    PAGE_KERNEL);
	if ( !entry->virtual ) {
		drm_free( entry->pagelist,
			  entry->pages * sizeof(*entry->pagelist),
			  DRM_MEM_PAGES );
		drm_free( entry,
			  sizeof(*entry),
			  DRM_MEM_SGLISTS );
		return -ENOMEM;
	}

	/* This also forces the mapping of COW pages, so our page list
	 * will be valid.  Please don't remove it...
	 */
	memset( entry->virtual, 0, pages << PAGE_SHIFT );

	entry->handle = (unsigned long)entry->virtual;

	DRM_DEBUG( "sg alloc handle  = %08lx\n", entry->handle );
	DRM_DEBUG( "sg alloc virtual = %p\n", entry->virtual );

	for ( i = entry->handle, j = 0 ; j < pages ; i += PAGE_SIZE, j++ ) {
		pgd = pgd_offset_k( i );
		pmd = pmd_offset( pgd, i );
		pte = pte_offset( pmd, i );

		entry->pagelist[j]= pte_page( *pte );
		SetPageReserved( entry->pagelist[j] );

		if ( j < 16 ) {
			DRM_DEBUG("0x%08lx (page %lu) => 0x%08lx\n",
				  i, j,
				  (unsigned long)entry->pagelist[j]->virtual);
		}
	}

	request.handle = entry->handle;

	if ( copy_to_user( (drm_scatter_gather_t *)arg,
			   &request,
			   sizeof(request) ) ) {
		drm_sg_cleanup( entry );
		return -EFAULT;
	}

	dev->sg = entry;

#ifdef DEBUG_SCATTER
	/* Verify that each page points to its virtual address, and vice
	 * versa.
	 */
	{
	int error = 0;

	for(i = 0; i < pages; i++) {
		unsigned long *tmp;

		tmp = (unsigned long *)entry->pagelist[i]->virtual;
		for(j = 0; j < PAGE_SIZE / sizeof(unsigned long); j++, tmp++) {
			*tmp = 0xcafebabe;
		}
		tmp = (unsigned long *)((u8 *)entry->virtual +
					(PAGE_SIZE * i));
		for(j = 0; j < PAGE_SIZE / sizeof(unsigned long); j++, tmp++) {
			if(*tmp != 0xcafebabe && error == 0) {
				error = 1;
				printk("Scatter allocation error, pagelist"
				       " does not match virtual mapping\n");
			}
		}
		tmp = (unsigned long *)entry->pagelist[i]->virtual;
		for(j = 0; j < PAGE_SIZE / sizeof(unsigned long); j++, tmp++) {
			*tmp = 0;
		}
	}
	if(error == 0) printk("Scatter allocation matches pagelist\n");
	}
#endif

	return 0;
}

int drm_sg_free( struct inode *inode, struct file *filp,
		 unsigned int cmd, unsigned long arg )
{
	drm_file_t *priv = filp->private_data;
	drm_device_t *dev = priv->dev;
	drm_scatter_gather_t request;
	drm_sg_mem_t *entry;

	if ( copy_from_user( &request,
			     (drm_scatter_gather_t *)arg,
			     sizeof(request) ) )
		return -EFAULT;

	entry = dev->sg;
	dev->sg = NULL;

	if ( !entry || entry->handle != request.handle ) return -EINVAL;

	DRM_INFO( "sg free virtual  = %p\n", entry->virtual );

	drm_sg_cleanup( entry );

	return 0;
}

  reply	other threads:[~2001-01-19 16:48 UTC|newest]

Thread overview: 55+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2001-01-19  3:26 PPC Lockup (ati-pcigart-branch) Michel Dänzer
2001-01-19  3:55 ` Dan Malek
2001-01-19  6:53   ` [Dri-devel] " Gareth Hughes
2001-01-19 16:48     ` Jeff Hartmann [this message]
2001-01-19 17:24     ` Dan Malek
2001-01-20  0:45       ` Gareth Hughes
2001-01-19 16:40 ` [Dri-devel] " Jeff Hartmann
2001-01-19 17:11   ` Benjamin Herrenschmidt
2001-01-19 22:26     ` Chris Emerson
2001-01-19 22:59       ` Benjamin Herrenschmidt
2001-01-19 23:43         ` Chris Emerson
2001-01-20  1:38           ` Benjamin Herrenschmidt
2001-01-20 13:21             ` Michael Schmitz
2001-01-20 16:00               ` Benjamin Herrenschmidt
2001-01-20 17:03                 ` Michael Schmitz
2001-01-20  2:46     ` Michel Dänzer
2001-01-20  4:17       ` Michel Dänzer
2001-01-22  9:44         ` Michel Dänzer
2001-01-22 17:59           ` Roman Zippel
2001-01-22 18:18             ` Michel Dänzer
2001-01-22 18:54               ` Roman Zippel
2001-01-22 19:39                 ` Dan Malek
2001-01-22 20:08                   ` Michel Dänzer
2001-01-22 20:30                   ` Jeff Hartmann
2001-01-22 21:23                     ` Roman Zippel
2001-01-22 23:12                       ` Frank Rowand
2001-01-22 21:31                     ` Dan Malek
2001-01-22 21:48                       ` Jeff Hartmann
2001-01-22 22:15                         ` Roman Zippel
2001-01-23 16:14                         ` Mike Beede
2001-01-22 22:31                       ` Roman Zippel
2001-01-23  0:24                         ` Dan Malek
2001-01-23  2:28                           ` Takashi Oe
2001-01-23  2:40                             ` Dan Malek
2001-01-23  4:40                               ` Ralph Metzler
2001-01-23  5:48                               ` Takashi Oe
2001-01-23 11:24                           ` Roman Zippel
2001-01-23  0:34                         ` Frank Rowand
2001-01-23  0:43                           ` Frank Rowand
2001-01-23 11:32                           ` Roman Zippel
2001-01-22 20:43                   ` Roman Zippel
2001-01-22 21:07                     ` Jeff Hartmann
2001-01-22 17:33         ` Dan Malek
2001-01-22 17:38           ` Jeff Hartmann
2001-01-22 17:38           ` Gareth Hughes
2001-01-22 17:43           ` Michel Dänzer
2001-01-22 18:36             ` Dan Malek
2001-01-22 18:44               ` Jeff Hartmann
2001-01-22 18:47               ` Michel Dänzer
2001-01-22 21:13         ` Dan Malek
2001-01-22 21:58           ` Roman Zippel
2001-01-22 23:48         ` Paul Mackerras
2001-01-23  0:13           ` Dan Malek
2001-01-20 13:15       ` Michael Schmitz
2001-01-19 17:11   ` Wolfgang Denk

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=3A686FF2.8000204@valinux.com \
    --to=jhartmann@valinux.com \
    --cc=dan@mvista.com \
    --cc=dri-devel@lists.sourceforge.net \
    --cc=gareth@valinux.com \
    --cc=linuxppc-dev@lists.linuxppc.org \
    --cc=michdaen@iiic.ethz.ch \
    /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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).