All of lore.kernel.org
 help / color / mirror / Atom feed
From: Marcin Slusarz <marcin.slusarz-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
To: Ben Skeggs <skeggsb-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
Cc: nouveau-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW@public.gmane.org
Subject: Re: [PATCH] drm/nouveau: fix early vram corruption originating from vgacon
Date: Thu, 4 Oct 2012 13:35:15 +0200	[thread overview]
Message-ID: <20121004113515.GC4979@joi.lan> (raw)
In-Reply-To: <20120912225230.GD8067-OI9uyE9O0yo@public.gmane.org>

On Thu, Sep 13, 2012 at 12:52:30AM +0200, Marcin Slusarz wrote:
> There's a short window between module load and fbcon initalization when
> it's possible for vgacon to write to VGA RAM. Nouveau uses this memory
> for different purposes, so if we are unlucky, it causes mysterious memory
> corruptions.
> 
> For me, booting with nv_printk debug levels set to 5 was enough to trigger it.
> It manifested as long stream of:
> "trapped write at ... on channel 0x0001fea0 BAR/PFIFO_WRITE/IN reason:
> DMAOBJ_LIMIT / PT_NOT_PRESENT / PAGE_SYSTEM_ONLY / PAGE_NOT_PRESENT"
> which eventually lead to complete hang.
> 
> Disabling access to VGA memory (through 0x54 PCI config space register) is
> enough to fix it, but it breaks copying screen data between old and new
> console (because old data is inaccessible). But blanking console (with
> entering_gfx==1) is enough to move vgacon screen buffer from VRAM to RAM
> and let handover to work correctly.
> 
> Signed-off-by: Marcin Slusarz <marcin.slusarz-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
> ---
>  drivers/gpu/drm/nouveau/nouveau_drm.c | 31 ++++++++++++++++++++++++++++++-
>  1 file changed, 30 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/gpu/drm/nouveau/nouveau_drm.c b/drivers/gpu/drm/nouveau/nouveau_drm.c
> index 6826525..1641bd9 100644
> --- a/drivers/gpu/drm/nouveau/nouveau_drm.c
> +++ b/drivers/gpu/drm/nouveau/nouveau_drm.c
> @@ -25,6 +25,7 @@
>  #include <linux/console.h>
>  #include <linux/module.h>
>  #include <linux/pci.h>
> +#include <linux/vt_kern.h>
>  
>  #include <core/device.h>
>  #include <core/client.h>
> @@ -51,6 +52,8 @@
>  
>  #include "nouveau_ttm.h"
>  
> +#define NV_PCI_VGAMEM_ENABLE 0x54
> +
>  MODULE_PARM_DESC(config, "option string to pass to driver core");
>  static char *nouveau_config;
>  module_param_named(config, nouveau_config, charp, 0400);
> @@ -247,9 +250,20 @@ nouveau_drm_load(struct drm_device *dev, unsigned long flags)
>  	struct nouveau_drm *drm;
>  	int ret;
>  
> +	/* Blank initial console to prevent VRAM corruption while we initialize
> +	 * the HW. For vgacon it will move console memory from VGA VRAM to RAM.
> +	 */
> +	console_lock();
> +	do_blank_screen(1);
> +	console_unlock();
> +
> +	/* Completely disable access to VGA IO/memory, just to be sure no one
> +	 * will change it. */
> +	pci_write_config_byte(pdev, NV_PCI_VGAMEM_ENABLE, 0);
> +
>  	ret = nouveau_cli_create(pdev, "DRM", sizeof(*drm), (void**)&drm);
>  	if (ret)
> -		return ret;
> +		goto fail_cli;
>  
>  	dev->dev_private = drm;
>  	drm->dev = dev;
> @@ -336,6 +350,11 @@ nouveau_drm_load(struct drm_device *dev, unsigned long flags)
>  
>  	nouveau_accel_init(drm);
>  	nouveau_fbcon_init(dev);
> +
> +	console_lock();
> +	do_unblank_screen(1);
> +	console_unlock();
> +
>  	return 0;
>  
>  fail_dispinit:
> @@ -351,12 +370,20 @@ fail_ttm:
>  	nouveau_vga_fini(drm);
>  fail_device:
>  	nouveau_cli_destroy(&drm->client);
> +fail_cli:
> +	pci_write_config_byte(pdev, NV_PCI_VGAMEM_ENABLE, 1);
> +
> +	console_lock();
> +	do_unblank_screen(1);
> +	console_unlock();
> +
>  	return ret;
>  }
>  
>  static int
>  nouveau_drm_unload(struct drm_device *dev)
>  {
> +	struct pci_dev *pdev = dev->pdev;
>  	struct nouveau_drm *drm = nouveau_drm(dev);
>  
>  	nouveau_fbcon_fini(dev);
> @@ -375,6 +402,8 @@ nouveau_drm_unload(struct drm_device *dev)
>  	nouveau_vga_fini(drm);
>  
>  	nouveau_cli_destroy(&drm->client);
> +
> +	pci_write_config_byte(pdev, NV_PCI_VGAMEM_ENABLE, 1);
>  	return 0;
>  }
>  
> -- 

What's up with this patch?

  parent reply	other threads:[~2012-10-04 11:35 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2012-09-12 22:52 [PATCH] drm/nouveau: fix early vram corruption originating from vgacon Marcin Slusarz
     [not found] ` <20120912225230.GD8067-OI9uyE9O0yo@public.gmane.org>
2012-10-04 11:35   ` Marcin Slusarz [this message]
     [not found]     ` <20121004113515.GC4979-OI9uyE9O0yo@public.gmane.org>
2012-10-04 11:44       ` Marcin Kościelnicki
     [not found]         ` <506D768D.6050100-mP9o5jsk0RY@public.gmane.org>
2012-10-05 20:18           ` Marcin Slusarz

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=20121004113515.GC4979@joi.lan \
    --to=marcin.slusarz-re5jqeeqqe8avxtiumwx3w@public.gmane.org \
    --cc=nouveau-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW@public.gmane.org \
    --cc=skeggsb-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.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.