Re: 2.6.12-rc1-mm2 -- nvidiafb driver gives black screen

Re: 2.6.12-rc1-mm2 -- nvidiafb driver gives black screen

[Andrew Morton]

(Added the mailing list - trust that's ok)

Miles Lane <miles.lane@gmail.com> wrote:
>
> Kernel command line: BOOT_IMAGE=Linux ro root=306 video=nvidiafb
> init=/etc/init lang=us apm=power-off nomce
> nvidiafb: nVidia device/chipset 10DE0312
> nvidiafb: nVidia Corporation NV31 [GeForce FX 5600]
> nvidiafb: CRTC0 found
> nvidiafb: CRTC1 not found
> nvidiafb: EDID found from BUS1
> nvidiafb: CRTC 0 appears to have a CRT attached
> nvidiafb: Using CRT on CRTC 0
> 
> >> I am wondering whether the following two lines point to the problem.
> 
> allocation failed: out of vmalloc space - use vmalloc=<size> to increase size.
> nvidiafb: cannot ioremap FB base

Quite possibly.

I'll do rc2-mm3 tonight - please test that.  It has fixes.

> CONFIG_AGP_NVIDIA=y
> CONFIG_I2C=y
> CONFIG_I2C_CHARDEV=m
> CONFIG_I2C_ALGOBIT=y
> CONFIG_I2C_ALGOPCF=m
> CONFIG_I2C_ALGOPCA=m
> CONFIG_I2C_ISA=m
> CONFIG_I2C_NFORCE2=m
> 
> CONFIG_FB=y
> CONFIG_FB_CFB_FILLRECT=y
> CONFIG_FB_CFB_COPYAREA=y
> CONFIG_FB_CFB_IMAGEBLIT=y
> CONFIG_FB_SOFT_CURSOR=y
> CONFIG_FB_MODE_HELPERS=y
> CONFIG_FB_TILEBLITTING=y
> CONFIG_VIDEO_SELECT=y
> CONFIG_FB_NVIDIA=y
> CONFIG_FB_NVIDIA_I2C=y
> CONFIG_FRAMEBUFFER_CONSOLE=y


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: 2.6.12-rc1-mm2 -- nvidiafb driver gives black screen

[Miles Lane]

On Thu, 24 Mar 2005 20:31:18 -0800, Andrew Morton <akpm@osdl.org> wrote:
> 
> (Added the mailing list - trust that's ok)
> 
> Miles Lane <miles.lane@gmail.com> wrote:
> >
> > Kernel command line: BOOT_IMAGE=Linux ro root=306 video=nvidiafb
> > init=/etc/init lang=us apm=power-off nomce
> > nvidiafb: nVidia device/chipset 10DE0312
> > nvidiafb: nVidia Corporation NV31 [GeForce FX 5600]
> > nvidiafb: CRTC0 found
> > nvidiafb: CRTC1 not found
> > nvidiafb: EDID found from BUS1
> > nvidiafb: CRTC 0 appears to have a CRT attached
> > nvidiafb: Using CRT on CRTC 0
> >
> > >> I am wondering whether the following two lines point to the problem.
> >
> > allocation failed: out of vmalloc space - use vmalloc=<size> to increase size.
> > nvidiafb: cannot ioremap FB base
> 
> Quite possibly.
> 
> I'll do rc2-mm3 tonight - please test that.  It has fixes.

Hello Andrew,

mm3 did not fix the problem.  I still get a blank framebuffer console.
I still get these two errors:
allocation failed: out of vmalloc space - use vmalloc=<size> to increase size.
nvidiafb: cannot ioremap FB base
Also, I still get the /sys listing Oops when listing the i2c entries
for the nvidiafb.
Lastly, I get the Oops that halts booting if I include certain other
i2c drivers in the build.
Removing nvidiafb from the build allows me to get a completely usable
build of 2.6.12-rc1-mm3.

I haven't heard anything from the author of the nvidiafb driver about
the various issues I have reported.

Thanks,
               Miles


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: Re: 2.6.12-rc1-mm2 -- nvidiafb driver gives black screen

[Randy.Dunlap]

Miles Lane wrote:
> On Thu, 24 Mar 2005 20:31:18 -0800, Andrew Morton <akpm@osdl.org> wrote:
> 
>>(Added the mailing list - trust that's ok)
>>
>>Miles Lane <miles.lane@gmail.com> wrote:
>>
>>>Kernel command line: BOOT_IMAGE=Linux ro root=306 video=nvidiafb
>>>init=/etc/init lang=us apm=power-off nomce
>>>nvidiafb: nVidia device/chipset 10DE0312
>>>nvidiafb: nVidia Corporation NV31 [GeForce FX 5600]
>>>nvidiafb: CRTC0 found
>>>nvidiafb: CRTC1 not found
>>>nvidiafb: EDID found from BUS1
>>>nvidiafb: CRTC 0 appears to have a CRT attached
>>>nvidiafb: Using CRT on CRTC 0
>>>
>>>
>>>>>I am wondering whether the following two lines point to the problem.
>>>
>>>allocation failed: out of vmalloc space - use vmalloc=<size> to increase size.
>>>nvidiafb: cannot ioremap FB base
>>
>>Quite possibly.
>>
>>I'll do rc2-mm3 tonight - please test that.  It has fixes.
> 
> 
> Hello Andrew,
> 
> mm3 did not fix the problem.  I still get a blank framebuffer console.
> I still get these two errors:
> allocation failed: out of vmalloc space - use vmalloc=<size> to increase size.
> nvidiafb: cannot ioremap FB base
> Also, I still get the /sys listing Oops when listing the i2c entries
> for the nvidiafb.
> Lastly, I get the Oops that halts booting if I include certain other
> i2c drivers in the build.
> Removing nvidiafb from the build allows me to get a completely usable
> build of 2.6.12-rc1-mm3.
> 
> I haven't heard anything from the author of the nvidiafb driver about
> the various issues I have reported.

I guess that there is some chance that the io_remap_pfn* patches
are involved with this.  Is there some way to determine exactly which
io remap kernel calls are involved in "nvidiafb: cannot ioremap FB 
base" ?  Is that in the exposed source code?  If so, where is that
source code?  If not, can we determine from the binary which calls
are being made?  Or should I just make a io_remap debug patch that
prints out lots of debug info so that we can track this down?


-- 
~Randy


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: Re: 2.6.12-rc1-mm2 -- nvidiafb driver gives black screen

[Miles Lane]

I am looking forward to having something to test.  Please keep me posted.

     Miles


-------------------------------------------------------
This SF.net email is sponsored by Demarc:
A global provider of Threat Management Solutions.
Download our HomeAdmin security software for free today!
http://www.demarc.com/info/Sentarus/hamr30

Re: Re: 2.6.12-rc1-mm2 -- nvidiafb driver gives black screen

[Randy.Dunlap]

Miles Lane wrote:
> I am looking forward to having something to test.  Please keep me posted.
> 
>      Miles

Well, you didn't answer any of my questions..... :(

and you dropped akpm from the cc: list, which is a nono.

I'll make some debug patch, but it would help if you answered my
questions.

-- 
~Randy


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: Re: 2.6.12-rc1-mm2 -- nvidiafb driver gives black screen

[Miles Lane]

On Mon, 28 Mar 2005 09:12:49 -0800, Randy.Dunlap <rddunlap@osdl.org> wrote:

By the why, I don't know why Andrew got dropped from my previous message list.
Also, I thought you were asking these questions of the fbdev and
vidiafb developers.

> I guess that there is some chance that the io_remap_pfn* patches
> are involved with this.  

I am a tester, so I don't have knowledge about this.

> Is there some way to determine exactly which
> io remap kernel calls are involved in "nvidiafb: cannot ioremap FB
> base" ?  

I don't know.  Someone who does, please tell me.  If someone
will tell me how, I'll be happy to gather the information.

> Is that in the exposed source code?  

I have no idea.  How would I find out?

> If so, where is that source code?  

I have no clue.

> If not, can we determine from the binary which calls
> are being made?  

Uhhh.  I have delivered the Oops info and a bunch of other
information.  I have given you whatever I thought might be helpful.  I
need guidance here.

> Or should I just make a io_remap debug patch that
> prints out lots of debug info so that we can track this down?

I thought you would determine this with input from developers.

Your humble tester (remember me from the early USB days?),

      Miles


-------------------------------------------------------
This SF.net email is sponsored by Demarc:
A global provider of Threat Management Solutions.
Download our HomeAdmin security software for free today!
http://www.demarc.com/info/Sentarus/hamr30

Re: Re: 2.6.12-rc1-mm2 -- nvidiafb driver gives black screen

[Randy.Dunlap]

Miles Lane wrote:
> On Mon, 28 Mar 2005 09:12:49 -0800, Randy.Dunlap <rddunlap@osdl.org> wrote:
> 
> By the why, I don't know why Andrew got dropped from my previous message list.
> Also, I thought you were asking these questions of the fbdev and
> vidiafb developers.
> 
> 
>>I guess that there is some chance that the io_remap_pfn* patches
>>are involved with this.  
> 
> 
> I am a tester, so I don't have knowledge about this.

Sure, it wasn't a question.  :)

> 
>>Is there some way to determine exactly which
>>io remap kernel calls are involved in "nvidiafb: cannot ioremap FB
>>base" ?  
> 
> 
> I don't know.  Someone who does, please tell me.  If someone
> will tell me how, I'll be happy to gather the information.
 >
> 
>>Is that in the exposed source code?  
> 
> 
> I have no idea.  How would I find out?
> 
>>If so, where is that source code?  
> 
> 
> I have no clue.

Where did you get the nvidia wrapper?
Where can I find it?

Where did you get the nvidia binary blob?
Where can I get it?

>>If not, can we determine from the binary which calls
>>are being made?  
> 
> 
> Uhhh.  I have delivered the Oops info and a bunch of other
> information.  I have given you whatever I thought might be helpful.  I
> need guidance here.

Sure, but the part that I really need to see is likely partially 
hidden in some binary blob.  We'll see.

> 
>>Or should I just make a io_remap debug patch that
>>prints out lots of debug info so that we can track this down?
> 
> 
> I thought you would determine this with input from developers.
> 
> Your humble tester (remember me from the early USB days?),

Sure I do.  :)
I'll make a patch, probably tonight Pacific time.

-- 
~Randy


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: Re: 2.6.12-rc1-mm2 -- nvidiafb driver gives black screen

[Miles Lane]

On Mon, 28 Mar 2005 17:02:38 -0800, Randy.Dunlap <rddunlap@osdl.org> wrote:
> Miles Lane wrote:
> > On Mon, 28 Mar 2005 09:12:49 -0800, Randy.Dunlap <rddunlap@osdl.org> wrote:
> >
> > By the why, I don't know why Andrew got dropped from my previous message list.
> > Also, I thought you were asking these questions of the fbdev and
> > vidiafb developers.
> >
> >
> >>I guess that there is some chance that the io_remap_pfn* patches
> >>are involved with this.
> >
> >
> > I am a tester, so I don't have knowledge about this.
> 
> Sure, it wasn't a question.  :)
> 
> >
> >>Is there some way to determine exactly which
> >>io remap kernel calls are involved in "nvidiafb: cannot ioremap FB
> >>base" ?
> >
> >
> > I don't know.  Someone who does, please tell me.  If someone
> > will tell me how, I'll be happy to gather the information.
>  >
> >
> >>Is that in the exposed source code?
> >
> >
> > I have no idea.  How would I find out?
> >
> >>If so, where is that source code?
> >
> >
> > I have no clue.
> 
> Where did you get the nvidia wrapper?
> Where can I find it?

Perhaps you are thinking I am using the NVidia binary driver?
This is not the case.  nvidiafb is a recently added framebuffer
driver (IIRC, first showing up in 2.6.12-rc1-mm2).  I believe
khali@linux-fr.org is the author.  I have reproduced this 
problem with both mm2 and mm3.

Do you need my .config file?

> Where did you get the nvidia binary blob?

If you had my .config file, couldn't you build the driver into
the kernel, as I have?  I don't know what a driver blob is.
I haven't built the driver as a module.

> Where can I get it?

Build it?

> >>If not, can we determine from the binary which calls
> >>are being made?
> >
> >
> > Uhhh.  I have delivered the Oops info and a bunch of other
> > information.  I have given you whatever I thought might be helpful.  I
> > need guidance here.
> 
> Sure, but the part that I really need to see is likely partially
> hidden in some binary blob.  We'll see.

Blob? Well, I have some i2c built as modules.

> > Your humble tester (remember me from the early USB days?),
> 
> Sure I do.  :)
> I'll make a patch, probably tonight Pacific time.

Great!
      Miles


-------------------------------------------------------
This SF.net email is sponsored by Demarc:
A global provider of Threat Management Solutions.
Download our HomeAdmin security software for free today!
http://www.demarc.com/info/Sentarus/hamr30

Re: Re: 2.6.12-rc1-mm2 -- nvidiafb driver gives black screen

[Randy.Dunlap]

Miles Lane wrote:
> On Mon, 28 Mar 2005 17:02:38 -0800, Randy.Dunlap <rddunlap@osdl.org> wrote:
> 
>>Miles Lane wrote:
>>
>>>On Mon, 28 Mar 2005 09:12:49 -0800, Randy.Dunlap <rddunlap@osdl.org> wrote:
>>>
>>>By the why, I don't know why Andrew got dropped from my previous message list.
>>>Also, I thought you were asking these questions of the fbdev and
>>>vidiafb developers.
>>>
>>>
>>>
>>>>I guess that there is some chance that the io_remap_pfn* patches
>>>>are involved with this.
>>>
>>>
>>>I am a tester, so I don't have knowledge about this.
>>
>>Sure, it wasn't a question.  :)
>>
>>
>>>>Is there some way to determine exactly which
>>>>io remap kernel calls are involved in "nvidiafb: cannot ioremap FB
>>>>base" ?
>>>
>>>
>>>I don't know.  Someone who does, please tell me.  If someone
>>>will tell me how, I'll be happy to gather the information.
>>
>> >
>>
>>>>Is that in the exposed source code?
>>>
>>>
>>>I have no idea.  How would I find out?
>>>
>>>
>>>>If so, where is that source code?
>>>
>>>
>>>I have no clue.
>>
>>Where did you get the nvidia wrapper?
>>Where can I find it?
> 
> 
> Perhaps you are thinking I am using the NVidia binary driver?
> This is not the case.  nvidiafb is a recently added framebuffer
> driver (IIRC, first showing up in 2.6.12-rc1-mm2).  I believe
> khali@linux-fr.org is the author.  I have reproduced this 
> problem with both mm2 and mm3.
> 
> Do you need my .config file?

I'll let you know.
I wondered while driving home if you were using the
(new) in-kernel nv driver.  Thanks for you and Andrew
confirming that.  Looks like I didn't go back far enough
in the thread history.

> 
>>Where did you get the nvidia binary blob?
> 
> 
> If you had my .config file, couldn't you build the driver into
> the kernel, as I have?  I don't know what a driver blob is.
> I haven't built the driver as a module.
> 
> 
>>Where can I get it?
> 
> 
> Build it?
> 
> 
>>>>If not, can we determine from the binary which calls
>>>>are being made?
>>>
>>>
>>>Uhhh.  I have delivered the Oops info and a bunch of other
>>>information.  I have given you whatever I thought might be helpful.  I
>>>need guidance here.
>>
>>Sure, but the part that I really need to see is likely partially
>>hidden in some binary blob.  We'll see.
> 
> 
> Blob? Well, I have some i2c built as modules.
> 
> 
>>>Your humble tester (remember me from the early USB days?),
>>
>>Sure I do.  :)
>>I'll make a patch, probably tonight Pacific time.

I send you something later tonight.

-- 
~Randy


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: Re: 2.6.12-rc1-mm2 -- nvidiafb driver gives black screen

[Jean Delvare]

Hi gentle folks,

> Perhaps you are thinking I am using the NVidia binary driver?
> This is not the case.  nvidiafb is a recently added framebuffer
> driver (IIRC, first showing up in 2.6.12-rc1-mm2).  I believe
> khali@linux-fr.org is the author.

I am not! Antonino Daplas is. Adding him to the CC list, although I guess
he reads linux-fbdev-devel.

The only part I can help with is the i2c stuff, which caused trouble
already, but so far I couldn't find anything relevant. The nvidiafb
driver is quite new, and it's natural that it needs some debugging. I
suspect that the driver has some severe problem with the way it handles
resources, as was suggested by our previous (failed) attempts to debug
it. I'd be happy to help with the debugging but unfortunately (no quite
so) I don't have any nvidia device in my machines anymore.

Somewhere else in this thread, Randy suggested that some __devinit stuff
wasn't really. I think it would be worth investigating this first, and
make sure that nothing is incorrectly tagged. The oops reported by Miles
sounded like large memory chunks had been freed when they shouldn't
have. I think I understand that erroneously tagging things __init might
lead to that kind of trouble. But maybe it's a completely different
problem, I'm really only suggesting this because I don't get what's
going on either.

Thanks,
--
Jean Delvare


-------------------------------------------------------
This SF.net email is sponsored by Demarc:
A global provider of Threat Management Solutions.
Download our HomeAdmin security software for free today!
http://www.demarc.com/info/Sentarus/hamr30

Re: Re: 2.6.12-rc1-mm2 -- nvidiafb driver gives black screen

[Andrew Morton]

"Randy.Dunlap" <rddunlap@osdl.org> wrote:
>
> >>If so, where is that source code?  
>  > 
>  > 
>  > I have no clue.
> 
>  Where did you get the nvidia wrapper?
>  Where can I find it?
> 
>  Where did you get the nvidia binary blob?
>  Where can I get it?

Miles is using drivers/video/nvidia/nvidiafb.o


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: Re: 2.6.12-rc1-mm2 -- nvidiafb driver gives black screen

[Miles Lane]

Here is some of the background on this problem:

http://marc.theaimsgroup.com/?l=linux-kernel&m=111076667232062&w=2
http://marc.theaimsgroup.com/?t=111172432600001&r=1&w=4
http://marc.theaimsgroup.com/?l=linux-kernel&m=111145286407616&w=2


-------------------------------------------------------
This SF.net email is sponsored by Demarc:
A global provider of Threat Management Solutions.
Download our HomeAdmin security software for free today!
http://www.demarc.com/info/Sentarus/hamr30

Re: Re: 2.6.12-rc1-mm2 -- nvidiafb driver gives black screen

[Randy.Dunlap]

[-- Attachment #1: Type: text/plain, Size: 2002 bytes --]

Miles Lane wrote:
> On Thu, 24 Mar 2005 20:31:18 -0800, Andrew Morton <akpm@osdl.org> wrote:
> 
>>(Added the mailing list - trust that's ok)
>>
>>Miles Lane <miles.lane@gmail.com> wrote:
>>
>>>Kernel command line: BOOT_IMAGE=Linux ro root=306 video=nvidiafb
>>>init=/etc/init lang=us apm=power-off nomce
>>>nvidiafb: nVidia device/chipset 10DE0312
>>>nvidiafb: nVidia Corporation NV31 [GeForce FX 5600]
>>>nvidiafb: CRTC0 found
>>>nvidiafb: CRTC1 not found
>>>nvidiafb: EDID found from BUS1
>>>nvidiafb: CRTC 0 appears to have a CRT attached
>>>nvidiafb: Using CRT on CRTC 0
>>>
>>>
>>>>>I am wondering whether the following two lines point to the problem.
>>>
>>>allocation failed: out of vmalloc space - use vmalloc=<size> to increase size.
>>>nvidiafb: cannot ioremap FB base
>>
>>Quite possibly.

Sure, without the FB memory, video won't show up.
And the 2 lines are directly related -- same error printed from
2 different places.

>>I'll do rc2-mm3 tonight - please test that.  It has fixes.
> 
> 
> Hello Andrew,
> 
> mm3 did not fix the problem.  I still get a blank framebuffer console.
> I still get these two errors:
> allocation failed: out of vmalloc space - use vmalloc=<size> to increase size.
> nvidiafb: cannot ioremap FB base
> Also, I still get the /sys listing Oops when listing the i2c entries
> for the nvidiafb.
> Lastly, I get the Oops that halts booting if I include certain other
> i2c drivers in the build.
> Removing nvidiafb from the build allows me to get a completely usable
> build of 2.6.12-rc1-mm3.
> 
> I haven't heard anything from the author of the nvidiafb driver about
> the various issues I have reported.

Here's a tiny patch to get started.  Hopefully it will show us
why ioremap is failing...

Another issue while looking at this code:
there is some __initdata in nvidia.c that should be __devinitdata
(deja vu)

Miles, after using this patch, please also test by booting
with "vmalloc=512M"  (or "vmalloc=256M" if you don't have 512 MB
to spare).

-- 
~Randy

[-- Attachment #2: nvid_remap_test1.patch --]
[-- Type: text/x-patch, Size: 934 bytes --]

linux-2612-rc1-mm3
Add a little debug printk to try to find out why ioremap fails.

Signed-off-by: Randy Dunlap <rddunlap@osdl.org>

diffstat:=
 drivers/video/nvidia/nvidia.c |    2 ++
 1 files changed, 2 insertions(+)

diff -Naurp ./drivers/video/nvidia/nvidia.c~nv_test_remap ./drivers/video/nvidia/nvidia.c
--- ./drivers/video/nvidia/nvidia.c~nv_test_remap	2005-03-28 19:21:28.000000000 -0800
+++ ./drivers/video/nvidia/nvidia.c	2005-03-28 19:46:47.000000000 -0800
@@ -1511,6 +1511,8 @@ static int __devinit nvidiafb_probe(stru
 	par->ScratchBufferSize = (par->Architecture < NV_ARCH_10) ? 8 * 1024 :
 	    16 * 1024;
 	par->ScratchBufferStart = par->FbUsableSize - par->ScratchBufferSize;
+	printk(KERN_ERR PFX "calling ioremap(smem_start = 0x%p, size = 0x%lx)\n",
+			nvidiafb_fix.smem_start, par->FbMapSize);
 	info->screen_base = ioremap(nvidiafb_fix.smem_start, par->FbMapSize);
 	nvidiafb_fix.smem_len = par->FbUsableSize;
 

[PATCH] nvidiafb: fix section references

[Randy.Dunlap]

[-- Attachment #1: Type: text/plain, Size: 2264 bytes --]

Randy.Dunlap wrote:
> Miles Lane wrote:
> 
>> On Thu, 24 Mar 2005 20:31:18 -0800, Andrew Morton <akpm@osdl.org> wrote:
>>
>>> (Added the mailing list - trust that's ok)
>>>
>>> Miles Lane <miles.lane@gmail.com> wrote:
>>>
>>>> Kernel command line: BOOT_IMAGE=Linux ro root=306 video=nvidiafb
>>>> init=/etc/init lang=us apm=power-off nomce
>>>> nvidiafb: nVidia device/chipset 10DE0312
>>>> nvidiafb: nVidia Corporation NV31 [GeForce FX 5600]
>>>> nvidiafb: CRTC0 found
>>>> nvidiafb: CRTC1 not found
>>>> nvidiafb: EDID found from BUS1
>>>> nvidiafb: CRTC 0 appears to have a CRT attached
>>>> nvidiafb: Using CRT on CRTC 0
>>>>
>>>>
>>>>>> I am wondering whether the following two lines point to the problem.
>>>>
>>>>
>>>> allocation failed: out of vmalloc space - use vmalloc=<size> to 
>>>> increase size.
>>>> nvidiafb: cannot ioremap FB base
>>>
>>>
>>> Quite possibly.
> 
> 
> Sure, without the FB memory, video won't show up.
> And the 2 lines are directly related -- same error printed from
> 2 different places.
> 
>>> I'll do rc2-mm3 tonight - please test that.  It has fixes.
>>
>>
>>
>> Hello Andrew,
>>
>> mm3 did not fix the problem.  I still get a blank framebuffer console.
>> I still get these two errors:
>> allocation failed: out of vmalloc space - use vmalloc=<size> to 
>> increase size.
>> nvidiafb: cannot ioremap FB base
>> Also, I still get the /sys listing Oops when listing the i2c entries
>> for the nvidiafb.
>> Lastly, I get the Oops that halts booting if I include certain other
>> i2c drivers in the build.
>> Removing nvidiafb from the build allows me to get a completely usable
>> build of 2.6.12-rc1-mm3.
>>
>> I haven't heard anything from the author of the nvidiafb driver about
>> the various issues I have reported.
> 
> 
> Here's a tiny patch to get started.  Hopefully it will show us
> why ioremap is failing...
> 
> Another issue while looking at this code:
> there is some __initdata in nvidia.c that should be __devinitdata
> (deja vu)
> 
> Miles, after using this patch, please also test by booting
> with "vmalloc=512M"  (or "vmalloc=256M" if you don't have 512 MB
> to spare).

and here are the section usage fixes.
Miles, please apply and test with this also.

Andrew, I think you and Tony want this.

-- 
~Randy

[-- Attachment #2: nvid_sections.patch --]
[-- Type: text/x-patch, Size: 2026 bytes --]

linux-2612-rc1-mm3
Fix section usage in nvidiafb driver.

Signed-off-by: Randy Dunlap <rddunlap@osdl.org>

diffstat:=
 drivers/video/nvidia/nvidia.c |   22 +++++++++++-----------
 1 files changed, 11 insertions(+), 11 deletions(-)

diff -Naurp ./drivers/video/nvidia/nvidia.c~nvid_sections ./drivers/video/nvidia/nvidia.c
--- ./drivers/video/nvidia/nvidia.c~nvid_sections	2005-03-28 08:40:09.000000000 -0800
+++ ./drivers/video/nvidia/nvidia.c	2005-03-29 08:20:20.000000000 -0800
@@ -402,25 +402,25 @@ static struct pci_device_id nvidiafb_pci
 MODULE_DEVICE_TABLE(pci, nvidiafb_pci_tbl);
 
 /* command line data, set in nvidiafb_setup() */
-static int flatpanel __initdata = -1;	/* Autodetect later */
-static int forceCRTC __initdata = -1;
-static int hwcur __initdata = 0;
-static int noaccel __initdata = 0;
-static int noscale __initdata = 0;
-static int paneltweak __initdata = 0;
+static int flatpanel __devinitdata = -1;	/* Autodetect later */
+static int forceCRTC __devinitdata = -1;
+static int hwcur __devinitdata = 0;
+static int noaccel __devinitdata = 0;
+static int noscale __devinitdata = 0;
+static int paneltweak __devinitdata = 0;
 #ifdef CONFIG_MTRR
-static int nomtrr __initdata = 0;
+static int nomtrr __devinitdata = 0;
 #endif
 
-static char *mode_option __initdata = NULL;
+static char *mode_option __devinitdata = NULL;
 
-static struct fb_fix_screeninfo __initdata nvidiafb_fix = {
+static struct fb_fix_screeninfo __devinitdata nvidiafb_fix = {
 	.type = FB_TYPE_PACKED_PIXELS,
 	.xpanstep = 8,
 	.ypanstep = 1,
 };
 
-static struct fb_var_screeninfo __initdata nvidiafb_default_var = {
+static struct fb_var_screeninfo __devinitdata nvidiafb_default_var = {
 	.xres = 640,
 	.yres = 480,
 	.xres_virtual = 640,
@@ -1617,7 +1617,7 @@ static void __exit nvidiafb_remove(struc
  * ------------------------------------------------------------------------- */
 
 #ifndef MODULE
-static int __init nvidiafb_setup(char *options)
+static int __devinit nvidiafb_setup(char *options)
 {
 	char *this_opt;
 

Re: Re: 2.6.12-rc1-mm2 -- nvidiafb driver gives black screen

[Antonino A. Daplas]

On Friday 25 March 2005 12:31, Andrew Morton wrote:
> (Added the mailing list - trust that's ok)
>
> Miles Lane <miles.lane@gmail.com> wrote:
> > Kernel command line: BOOT_IMAGE=Linux ro root=306 video=nvidiafb
> > init=/etc/init lang=us apm=power-off nomce
> > nvidiafb: nVidia device/chipset 10DE0312
> > nvidiafb: nVidia Corporation NV31 [GeForce FX 5600]
> > nvidiafb: CRTC0 found
> > nvidiafb: CRTC1 not found
> > nvidiafb: EDID found from BUS1
> > nvidiafb: CRTC 0 appears to have a CRT attached
> > nvidiafb: Using CRT on CRTC 0
> >
> > >> I am wondering whether the following two lines point to the problem.
> >
> > allocation failed: out of vmalloc space - use vmalloc=<size> to increase
> > size. nvidiafb: cannot ioremap FB base
>

Sorry for answering very, very late. I have been busy for the past 2 weeks
and still am busy.

I'm attaching a patch that adds a vram option that allows the user to
specify the amount of video ram to remap.  This is useful for cards
with huge amounts of vram.

You can use, for example:

video=nvidiafb:vram:8

to remap only 8 MiB of video RAM.

(Or perhaps, we can always limit the amount to not more than 128 MiB? I'm 
attaching an incremental patch in another mail, if people want this).

As for the i2c problem, this is probably due to a misplaced label in the
failure path, and the reason why I had a difficult time reproducing this bug.
When nvidiafb failed with the ioremap, it skipped deletion of the i2c busses
causing the crash whenever the bus created by nvidiafb is read.  The fix is
also included in this patch.

So, can you test this patch and see if it helps?

From: Antonino Daplas <adaplas@pol.net>
Signed-off-by: Antonino Daplas <adaplas@pol.net>
---

 nvidia.c |   13 ++++++++++++-
 1 files changed, 12 insertions(+), 1 deletion(-)

diff -Nru a/drivers/video/nvidia/nvidia.c b/drivers/video/nvidia/nvidia.c
--- a/drivers/video/nvidia/nvidia.c	2005-04-14 12:44:06 +08:00
+++ b/drivers/video/nvidia/nvidia.c	2005-04-14 13:33:38 +08:00
@@ -408,6 +408,7 @@
 static int noaccel __devinitdata = 0;
 static int noscale __devinitdata = 0;
 static int paneltweak __devinitdata = 0;
+static int vram __devinitdata = 0;
 #ifdef CONFIG_MTRR
 static int nomtrr __devinitdata = 0;
 #endif
@@ -1507,6 +1508,10 @@
 
 	par->FbAddress = nvidiafb_fix.smem_start;
 	par->FbMapSize = par->RamAmountKBytes * 1024;
+
+	if (vram && vram * 1024 * 1024 < par->FbMapSize)
+	    par->FbMapSize = vram * 1024 * 1024;
+
 	par->FbUsableSize = par->FbMapSize - (128 * 1024);
 	par->ScratchBufferSize = (par->Architecture < NV_ARCH_10) ? 8 * 1024 :
 	    16 * 1024;
@@ -1566,9 +1571,9 @@
 
       err_out_iounmap_fb:
 	iounmap(info->screen_base);
+      err_out_free_base1:
 	fb_destroy_modedb(info->monspecs.modedb);
 	nvidia_delete_i2c_busses(par);
-      err_out_free_base1:
 	iounmap(par->REGS);
       err_out_free_base0:
 	pci_release_regions(pd);
@@ -1645,6 +1650,8 @@
 			noscale = 1;
 		} else if (!strncmp(this_opt, "paneltweak:", 11)) {
 			paneltweak = simple_strtoul(this_opt+11, NULL, 0);
+		} else if (!strncmp(this_opt, "vram:", 5)) {
+			vram = simple_strtoul(this_opt+5, NULL, 0);
 #ifdef CONFIG_MTRR
 		} else if (!strncmp(this_opt, "nomtrr", 6)) {
 			nomtrr = 1;
@@ -1716,6 +1723,10 @@
 MODULE_PARM_DESC(forceCRTC,
 		 "Forces usage of a particular CRTC in case autodetection "
 		 "fails. (0 or 1) (default=autodetect)");
+module_param(vram, int, 0);
+MODULE_PARM_DESC(vram,
+		 "amount of framebuffer memory to remap in MiB"
+		 "(default=0 - remap entire memory)");
 #ifdef CONFIG_MTRR
 module_param(nomtrr, bool, 0);
 MODULE_PARM_DESC(nomtrr, "Disables MTRR support (0 or 1=disabled) "





-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: Re: 2.6.12-rc1-mm2 -- nvidiafb driver gives black screen

[Jean Delvare]

Hi Tony,

> As for the i2c problem, this is probably due to a misplaced label in
> the failure path, and the reason why I had a difficult time
> reproducing this bug. When nvidiafb failed with the ioremap, it
> skipped deletion of the i2c busses causing the crash whenever the bus
> created by nvidiafb is read.  The fix is also included in this patch.

Exactly what I suspected from the very beginning, but I did not
investigate the error path back then - in fact I can't remember Miles
telling me that the the driver was actually not working for him. Thanks
a lot for confirming and fixing the problem.

> So, can you test this patch and see if it helps?

I can't - I have no compatible hardware - but hopefully Miles will be
able to test.

-- 
Jean Delvare


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: Re: 2.6.12-rc1-mm2 -- nvidiafb driver gives black screen

[Antonino A. Daplas]

On Friday 15 April 2005 01:06, Jean Delvare wrote:
> Hi Tony,
>
> > As for the i2c problem, this is probably due to a misplaced label in
> > the failure path, and the reason why I had a difficult time
> > reproducing this bug. When nvidiafb failed with the ioremap, it
> > skipped deletion of the i2c busses causing the crash whenever the bus
> > created by nvidiafb is read.  The fix is also included in this patch.
>
> Exactly what I suspected from the very beginning, but I did not
> investigate the error path back then - in fact I can't remember Miles
> telling me that the the driver was actually not working for him. Thanks

He didn't.  I just assumed that it did because he also reported that nvidiafb
failed to load due to an ioremap problem (too much video RAM).

> a lot for confirming and fixing the problem.

Well, not yet confirmed, I'll wait for Miles to do so.

Tony




-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: Re: 2.6.12-rc1-mm2 -- nvidiafb driver gives black screen

[Antonino A. Daplas]

On Friday 25 March 2005 12:31, Andrew Morton wrote:

> >
> > allocation failed: out of vmalloc space - use vmalloc=<size> to increase
> > size. nvidiafb: cannot ioremap FB base
>

This patch will always limit the amount of video RAM to remap at
128 MiB.

From: Antonino Daplas <adaplas@pol.net>
Signed-off-by: Antonino Daplas <adaplas@pol.net>
---

 nvidia.c |    6 +++++-
 1 files changed, 5 insertions(+), 1 deletion(-)

diff -Nru a/drivers/video/nvidia/nvidia.c b/drivers/video/nvidia/nvidia.c
--- a/drivers/video/nvidia/nvidia.c	2005-04-14 13:33:38 +08:00
+++ b/drivers/video/nvidia/nvidia.c	2005-04-14 13:44:15 +08:00
@@ -1510,7 +1510,11 @@
 	par->FbMapSize = par->RamAmountKBytes * 1024;
 
 	if (vram && vram * 1024 * 1024 < par->FbMapSize)
-	    par->FbMapSize = vram * 1024 * 1024;
+		par->FbMapSize = vram * 1024 * 1024;
+
+	/* Limit amount of vram to 128 MB */
+	if (par->FbMapSize > 128 * 1024 * 1024)
+		par->FbMapSize = 128 * 1024 * 1024;
 
 	par->FbUsableSize = par->FbMapSize - (128 * 1024);
 	par->ScratchBufferSize = (par->Architecture < NV_ARCH_10) ? 8 * 1024 :






-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: Re: 2.6.12-rc1-mm2 -- nvidiafb driver gives black screen

[Jean Delvare]

Hi Tony,

> This patch will always limit the amount of video RAM to remap at
> 128 MiB.

I don't much get the idea. What are you trying to prevent?

128 MiB seems a rather low limit. Graphics adapters with 256 MiB, 512
MiB or even 640 MiB exist, and I see little reason why it would stop
increasing.

> diff -Nru a/drivers/video/nvidia/nvidia.c b/drivers/video/nvidia/nvidia.c
> --- a/drivers/video/nvidia/nvidia.c	2005-04-14 13:33:38 +08:00
> +++ b/drivers/video/nvidia/nvidia.c	2005-04-14 13:44:15 +08:00
> @@ -1510,7 +1510,11 @@
>  	par->FbMapSize = par->RamAmountKBytes * 1024;
>  
>  	if (vram && vram * 1024 * 1024 < par->FbMapSize)
> -	    par->FbMapSize = vram * 1024 * 1024;
> +		par->FbMapSize = vram * 1024 * 1024;

You're fixing an indentation error you just introduced in the previous
patch. Maybe you could fix the first patch instead?

Thanks,
-- 
Jean Delvare


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: Re: 2.6.12-rc1-mm2 -- nvidiafb driver gives black screen

[Miles Lane]

Perhaps I am mistaken, but I thought Andrew wanted to see the nvidiafb
driver work with a default X86 kernel configuration (which allows for
up to 128M to be allocated).  Is the driver attempting to allocate
exactly 128M because that is how much video ram is present in my video
card?

If lots of video cards exist that have more than 128M of video ram,
won't we be seeing many boot problems with the default kernel
configuration?  Does this point to a problem in the kernel design?  Is
this limitation in the kernel's default configuration advantageous?

Thanks,
       Miles


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_ide95&alloc_id\x14396&op=click

Re: Re: 2.6.12-rc1-mm2 -- nvidiafb driver gives black screen

[Antonino A. Daplas]

On Friday 15 April 2005 02:19, Miles Lane wrote:
> Perhaps I am mistaken, but I thought Andrew wanted to see the nvidiafb
> driver work with a default X86 kernel configuration (which allows for
> up to 128M to be allocated).  Is the driver attempting to allocate
> exactly 128M because that is how much video ram is present in my video
> card?
>

Not allocation, as standalone cards have already a fixed amount of RAM. The
patch will limit the amount of video RAM that the kernel can access to a
maximum of 128 MiB.

> If lots of video cards exist that have more than 128M of video ram,
> won't we be seeing many boot problems with the default kernel
> configuration?  Does this point to a problem in the kernel design?  Is
> this limitation in the kernel's default configuration advantageous?
>

The patch is a short-term workaround where cards with large amounts
of video RAM eat a large amount of the kernel's address space.

What I intend to do is to mimic what radeonfb is doing. Ie, info->screen_size
points to the actual vram, while fix->smem_len points to ioremapped area only.
I'll do some testing first since nvidiafb uses the last chunk of the ioremapped
vram for the DMA buffer, and I don't know the consequence if this buffer is
unintentionally exported to user space, from the security point of view.

Tony




-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: Re: 2.6.12-rc1-mm2 -- nvidiafb driver gives black screen

[Antonino A. Daplas]

On Friday 15 April 2005 01:13, Jean Delvare wrote:
> Hi Tony,
>
> > This patch will always limit the amount of video RAM to remap at
> > 128 MiB.
>
> I don't much get the idea. What are you trying to prevent?

I'm preventing ioremap failures when the driver tries to remap a large
chunk of the graphics aperture.

>
> 128 MiB seems a rather low limit. Graphics adapters with 256 MiB, 512
> MiB or even 640 MiB exist, and I see little reason why it would stop
> increasing.

For fbcon use, such a large amount of RAM does not help significantly
with performance.  Also, nvidiafb cannot use the entire video RAM as
framebuffer space since it's 2D drawing functions are limited by the
clipping coordinates which is set at 0x7fff, x and y, ie, one can only do
blits within the range of 0x0-0x7fff.

Tony




-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: Re: 2.6.12-rc1-mm2 -- nvidiafb driver gives black screen

[Miles Lane]

Hello Tony,

So far no Oops messages while testing your patches to nvidiafb.
I have tried compiling all the I2C drivers into the kernel and then
booting with vesafb as the console graphics driver.  So far it looks
good.  I need to do more careful testing of the case where no
vmalloc= is specified.  I'll try to get to this tomorrow.  So far, so good.

                Miles


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_ide95&alloc_id\x14396&op=click

Re: Re: 2.6.12-rc1-mm2 -- nvidiafb driver gives black screen

[Miles Lane]

Your patches seem to have fixed the Oops with I2C, but 
nvidiafb is not working unless I specify more than vmalloc=128M
in the append options.  That is, the default kernel options
do not provide for a working nvidiafb display.

Here's what I get when testing 2.6.12-rc2-mm3
(I verified that your two patches have been applied in my
tree):

Kernel command line: BOOT_IMAGE=Linux ro root=306
video=nvidiafb:1280x1024-16@85 vmalloc=
128M init=/etc/init lang=us apm=power-off nomce

nvidiafb: nVidia device/chipset 10DE0312
nvidiafb: nVidia Corporation NV31 [GeForce FX 5600]
i2c_adapter i2c-0: registered as adapter #0
i2c_adapter i2c-1: registered as adapter #1
i2c_adapter i2c-2: registered as adapter #2
nvidiafb: CRTC0 found
nvidiafb: CRTC1 not found
i2c_adapter i2c-0: master_xfer[0] W, addr=0x50, len=1
i2c_adapter i2c-0: master_xfer[1] R, addr=0x50, len=128
nvidiafb: EDID found from BUS1
i2c_adapter i2c-1: master_xfer[0] W, addr=0x50, len=1
i2c_adapter i2c-1: master_xfer[1] R, addr=0x50, len=128
i2c_adapter i2c-1: master_xfer[0] W, addr=0x50, len=1
i2c_adapter i2c-1: master_xfer[1] R, addr=0x50, len=128
i2c_adapter i2c-1: master_xfer[0] W, addr=0x50, len=1
i2c_adapter i2c-1: master_xfer[1] R, addr=0x50, len=128
nvidiafb: CRTC 0 appears to have a CRT attached
nvidiafb: Using CRT on CRTC 0
allocation failed: out of vmalloc space - use vmalloc=<size> to increase size.
nvidiafb: cannot ioremap FB base

When I boot, the screen stays black until XFree86 starts up.


-------------------------------------------------------
This SF.Net email is sponsored by: New Crystal Reports XI.
Version 11 adds new functionality designed to reduce time involved in
creating, integrating, and deploying reporting solutions. Free runtime info,
new features, or free trial, at: http://www.businessobjects.com/devxi/728

Re: Re: 2.6.12-rc1-mm2 -- nvidiafb driver gives black screen

[Randy.Dunlap]

On Thu, 21 Apr 2005 01:51:12 -0400 Miles Lane wrote:

| Your patches seem to have fixed the Oops with I2C, but 
| nvidiafb is not working unless I specify more than vmalloc=128M
| in the append options.  That is, the default kernel options
| do not provide for a working nvidiafb display.
| 
| Here's what I get when testing 2.6.12-rc2-mm3
| (I verified that your two patches have been applied in my
| tree):
| 
| Kernel command line: BOOT_IMAGE=Linux ro root=306
| video=nvidiafb:1280x1024-16@85 vmalloc=
| 128M init=/etc/init lang=us apm=power-off nomce
| 
...
| allocation failed: out of vmalloc space - use vmalloc=<size> to increase size.
| nvidiafb: cannot ioremap FB base
| 
| When I boot, the screen stays black until XFree86 starts up.

Tony,

As I wrote to Miles a few weeks ago:
"I started looking at vmalloc() and what it calls (which is
__get_vm_area).  _get_vm_area() always allocates one extra
page (called a "guard page") between all vmalloc allocations,
so even though 128 MB is the default amount and the amount
that nvidiafb wants to use, the kernel wants to allocate
128 MB + PAGE_SIZE (4 KB on x86; are you on x86?), so even
if nvidiafb is the only caller, the vmalloc() call will
fail."

This means that the default kernel vmalloc size of 128MB
cannot be totally (fully) allocated due to this code:
	/*
	 * We always allocate a guard page.
	 */
	size += PAGE_SIZE;

IMO we (somebody) should add the guard page into the default
vmalloc size so that the full 128MB can be allocated.
I'll take a look at doing that... no promises.

---
~Randy


-------------------------------------------------------
This SF.Net email is sponsored by: New Crystal Reports XI.
Version 11 adds new functionality designed to reduce time involved in
creating, integrating, and deploying reporting solutions. Free runtime info,
new features, or free trial, at: http://www.businessobjects.com/devxi/728

Re: Re: 2.6.12-rc1-mm2 -- nvidiafb driver gives black screen

[Randy.Dunlap]

On Thu, 21 Apr 2005 09:22:15 -0700 Randy.Dunlap wrote:

| On Thu, 21 Apr 2005 01:51:12 -0400 Miles Lane wrote:
| 
| | Here's what I get when testing 2.6.12-rc2-mm3
| | (I verified that your two patches have been applied in my
| | tree):
| | 
| | Kernel command line: BOOT_IMAGE=Linux ro root=306
| | video=nvidiafb:1280x1024-16@85 vmalloc=
| | 128M init=/etc/init lang=us apm=power-off nomce
| | 
| ...
| | allocation failed: out of vmalloc space - use vmalloc=<size> to increase size.
| | nvidiafb: cannot ioremap FB base
| | 
| | When I boot, the screen stays black until XFree86 starts up.
| 
| Tony,
| 
| As I wrote to Miles a few weeks ago:
| "I started looking at vmalloc() and what it calls (which is
| __get_vm_area).  _get_vm_area() always allocates one extra
| page (called a "guard page") between all vmalloc allocations,
| so even though 128 MB is the default amount and the amount
| that nvidiafb wants to use, the kernel wants to allocate
| 128 MB + PAGE_SIZE (4 KB on x86; are you on x86?), so even
| if nvidiafb is the only caller, the vmalloc() call will
| fail."
| 
| This means that the default kernel vmalloc size of 128MB
| cannot be totally (fully) allocated due to this code:
| 	/*
| 	 * We always allocate a guard page.
| 	 */
| 	size += PAGE_SIZE;
| 
| IMO we (somebody) should add the guard page into the default
| vmalloc size so that the full 128MB can be allocated.
| I'll take a look at doing that... no promises.

Miles,

Can you test this patch, please?
(against 2.6.12-rc3 if it matters)

Don't use "vmalloc=size" on the kernel boot line.
The driver should still be able to allocate 128 MB
without using that.



From: Randy Dunlap <rddunlap@osdl.org>

Include a guard page in the default vmalloc reserved size so that
a vmalloc() of 128 MB will succeed.  Useful for framebuffer drivers
that vmalloc() 128 MB but that currently fails.
Specifying "vmalloc=size" on the kernel command line still allocates
exactly that requested size, with no extra for a guard page.

Signed-off-by: Randy Dunlap <rddunlap@osdl.org>
---

 arch/i386/mm/init.c |    3 ++-
 1 files changed, 2 insertions(+), 1 deletion(-)

diff -Naurp ./arch/i386/mm/init.c~vmalloc_guard ./arch/i386/mm/init.c
--- ./arch/i386/mm/init.c~vmalloc_guard	2005-04-21 08:15:35.000000000 -0700
+++ ./arch/i386/mm/init.c	2005-04-21 10:50:55.000000000 -0700
@@ -31,6 +31,7 @@
 #include <asm/processor.h>
 #include <asm/system.h>
 #include <asm/uaccess.h>
+#include <asm/page.h>
 #include <asm/pgtable.h>
 #include <asm/dma.h>
 #include <asm/fixmap.h>
@@ -40,7 +41,7 @@
 #include <asm/tlbflush.h>
 #include <asm/sections.h>
 
-unsigned int __VMALLOC_RESERVE = 128 << 20;
+unsigned int __VMALLOC_RESERVE = (128 << 20) + PAGE_SIZE;
 
 DEFINE_PER_CPU(struct mmu_gather, mmu_gathers);
 unsigned long highstart_pfn, highend_pfn;


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: Re: 2.6.12-rc1-mm2 -- nvidiafb driver gives black screen

[Miles Lane]

Hi Randy,

I tried your patch to a clean 2.6.12-rc2-mm3 tree (I backed out
Antonio's patches).  The result of not having a vmalloc= in the boot
arguments is that the screen is still black:

allocation failed: out of vmalloc space - use vmalloc=<size> to increase size.
nvidiafb: cannot ioremap FB base

I will try again with Antonio's patches added.

       Miles


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_ide95&alloc_id\x14396&op=click

Re: Re: 2.6.12-rc1-mm2 -- nvidiafb driver gives black screen

[Miles Lane]

Okay, it also fails with Antonio's patches applied in addition to yours.

Thanks,
           Miles

On 4/22/05, Miles Lane <miles.lane@gmail.com> wrote:
> Hi Randy,
> 
> I tried your patch to a clean 2.6.12-rc2-mm3 tree (I backed out
> Antonio's patches).  The result of not having a vmalloc= in the boot
> arguments is that the screen is still black:
> 
> allocation failed: out of vmalloc space - use vmalloc=<size> to increase size.
> nvidiafb: cannot ioremap FB base
> 
> I will try again with Antonio's patches added.
> 
>        Miles


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_ide95&alloc_id\x14396&op=click

Re: Re: 2.6.12-rc1-mm2 -- nvidiafb driver gives black screen

[Miles Lane]

On 4/22/05, Miles Lane <miles.lane@gmail.com> wrote:
> Okay, it also fails with Antonio's patches applied in addition to yours.

I know I implied this in my previous message, but I want to make it explicit.
This combination of patches also fails to give a working framebuffer
console when I specify vmalloc=128M.  I only get a working framebuffer
console by specifying vmalloc=256M.

           Miles


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_ide95&alloc_id\x14396&op=click

Re: Re: 2.6.12-rc1-mm2 -- nvidiafb driver gives black screen

[Antonino A. Daplas]

On Monday 25 April 2005 01:09, Miles Lane wrote:
> On 4/22/05, Miles Lane <miles.lane@gmail.com> wrote:
> > Okay, it also fails with Antonio's patches applied in addition to yours.
>
> I know I implied this in my previous message, but I want to make it
> explicit. This combination of patches also fails to give a working
> framebuffer console when I specify vmalloc=128M.  I only get a working
> framebuffer console by specifying vmalloc=256M.
>
>            Miles

Can you try this patch?  This limits the amount of RAM that nvidiafb
will remap to 64 MiB.  This is on top of another patch I sent to Andrew:
"[Linux-fbdev-devel] [PATCH 1/6] nvidiafb: ioremap and i2c fixes"

Tony

diff -Nru a/drivers/video/nvidia/nvidia.c b/drivers/video/nvidia/nvidia.c
--- a/drivers/video/nvidia/nvidia.c	2005-04-16 11:54:44 +08:00
+++ b/drivers/video/nvidia/nvidia.c	2005-04-25 10:39:45 +08:00
@@ -1511,9 +1511,9 @@
 	if (vram && vram * 1024 * 1024 < par->FbMapSize)
 		par->FbMapSize = vram * 1024 * 1024;
 
-	/* Limit amount of vram to 128 MB */
-	if (par->FbMapSize > 128 * 1024 * 1024)
-		par->FbMapSize = 128 * 1024 * 1024;
+	/* Limit amount of vram to 64 MB */
+	if (par->FbMapSize > 64 * 1024 * 1024)
+		par->FbMapSize = 64 * 1024 * 1024;
 
 	par->FbUsableSize = par->FbMapSize - (128 * 1024);
 	par->ScratchBufferSize = (par->Architecture < NV_ARCH_10) ? 8 * 1024 :




-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: Re: 2.6.12-rc1-mm2 -- nvidiafb driver gives black screen

[Miles Lane]

Yep, that seems to have fixed it.  

Thanks!
          Miles

On 4/24/05, Antonino A. Daplas <adaplas@hotpop.com> wrote:
>
> Can you try this patch?  This limits the amount of RAM that nvidiafb
> will remap to 64 MiB.  This is on top of another patch I sent to Andrew:
> "[Linux-fbdev-devel] [PATCH 1/6] nvidiafb: ioremap and i2c fixes"
> 
> Tony
> 
> diff -Nru a/drivers/video/nvidia/nvidia.c b/drivers/video/nvidia/nvidia.c
> --- a/drivers/video/nvidia/nvidia.c     2005-04-16 11:54:44 +08:00
> +++ b/drivers/video/nvidia/nvidia.c     2005-04-25 10:39:45 +08:00
> @@ -1511,9 +1511,9 @@
>         if (vram && vram * 1024 * 1024 < par->FbMapSize)
>                 par->FbMapSize = vram * 1024 * 1024;
> 
> -       /* Limit amount of vram to 128 MB */
> -       if (par->FbMapSize > 128 * 1024 * 1024)
> -               par->FbMapSize = 128 * 1024 * 1024;
> +       /* Limit amount of vram to 64 MB */
> +       if (par->FbMapSize > 64 * 1024 * 1024)
> +               par->FbMapSize = 64 * 1024 * 1024;
> 
>         par->FbUsableSize = par->FbMapSize - (128 * 1024);
>         par->ScratchBufferSize = (par->Architecture < NV_ARCH_10) ? 8 * 1024 :
> 
> -------------------------------------------------------
> SF email is sponsored by - The IT Product Guide
> Read honest & candid reviews on hundreds of IT Products from real users.
> Discover which products truly live up to the hype. Start reading now.
> http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
> _______________________________________________
> Linux-fbdev-devel mailing list
> Linux-fbdev-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/linux-fbdev-devel
>


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_ide95&alloc_id\x14396&op=click

Re: Re: 2.6.12-rc1-mm2 -- nvidiafb driver gives black screen

[Miles Lane]

Oops.  Sorry some of you will get this twice.

Yes, that patch fixed the problem.  I was able to get a usable
framebuffer console without specifying the vmalloc boot parameter.

Thanks!
          Miles

On 4/24/05, Antonino A. Daplas <adaplas@hotpop.com> wrote:
> 
> Can you try this patch?  This limits the amount of RAM that nvidiafb
> will remap to 64 MiB.  This is on top of another patch I sent to Andrew:
> "[Linux-fbdev-devel] [PATCH 1/6] nvidiafb: ioremap and i2c fixes"
> 
> Tony
> 
> diff -Nru a/drivers/video/nvidia/nvidia.c b/drivers/video/nvidia/nvidia.c
> --- a/drivers/video/nvidia/nvidia.c     2005-04-16 11:54:44 +08:00
> +++ b/drivers/video/nvidia/nvidia.c     2005-04-25 10:39:45 +08:00
> @@ -1511,9 +1511,9 @@
>         if (vram && vram * 1024 * 1024 < par->FbMapSize)
>                 par->FbMapSize = vram * 1024 * 1024;
> 
> -       /* Limit amount of vram to 128 MB */
> -       if (par->FbMapSize > 128 * 1024 * 1024)
> -               par->FbMapSize = 128 * 1024 * 1024;
> +       /* Limit amount of vram to 64 MB */
> +       if (par->FbMapSize > 64 * 1024 * 1024)
> +               par->FbMapSize = 64 * 1024 * 1024;
> 
>         par->FbUsableSize = par->FbMapSize - (128 * 1024);
>         par->ScratchBufferSize = (par->Architecture < NV_ARCH_10) ? 8 * 1024 :
> 
> -------------------------------------------------------
> SF email is sponsored by - The IT Product Guide
> Read honest & candid reviews on hundreds of IT Products from real users.
> Discover which products truly live up to the hype. Start reading now.
> http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
> _______________________________________________
> Linux-fbdev-devel mailing list
> Linux-fbdev-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/linux-fbdev-devel
>


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_ide95&alloc_id\x14396&op=click