All of lore.kernel.org
 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 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.