[arch-general] quiet kernel parameter
Does anybody know the difference between the quiet kernel parameter and loglevel=4 parameter? All of the info I can find states that they should be identical, and /proc/sys/kern -- Jason Steadman http://www.meyithi.com/ http://twitter.com/meyithi
Does anybody know the difference between the quiet kernel parameter and loglevel=4 parameter? All of the info I can find states that they should be identical, and /proc/sys/kernel/printk is identical. I am however getting framebuffer corruption during boot when using quiet which doesn't occur when using loglevel=4. -- Jason Steadman http://www.meyithi.com/ http://twitter.com/meyithi
Hi Jason, On Thu, Jul 28, 2011 at 5:16 PM, Meyithi <mail@meyithi.com> wrote:
Does anybody know the difference between the quiet kernel parameter and loglevel=4 parameter?
All of the info I can find states that they should be identical, and /proc/sys/kernel/printk is identical.
They should be identical. Just look at the code: <http://git.kernel.org/?p=linux/kernel/git/torvalds/linux.git;a=blob;f=init/main.c;h=d7211faed2adfb295caf46bbb9c70835f622eabe;hb=HEAD#l201>. The quiet parameter just sets loglevel=4. Unless I'm missing something.
I am however getting framebuffer corruption during boot when using quiet which doesn't occur when using loglevel=4.
This is very odd. Any more info on this? Screenshots? Can you see a difference in dmesg? Cheers, Tom
On 28 July 2011 16:55, Tom Gundersen <teg@jklm.no> wrote:
Hi Jason,
On Thu, Jul 28, 2011 at 5:16 PM, Meyithi <mail@meyithi.com> wrote:
Does anybody know the difference between the quiet kernel parameter and loglevel=4 parameter?
All of the info I can find states that they should be identical, and /proc/sys/kernel/printk is identical.
They should be identical. Just look at the code: < http://git.kernel.org/?p=linux/kernel/git/torvalds/linux.git;a=blob;f=init/m...
. The quiet parameter just sets loglevel=4. Unless I'm missing something.
I am however getting framebuffer corruption during boot when using quiet which doesn't occur when using loglevel=4.
This is very odd. Any more info on this? Screenshots? Can you see a difference in dmesg?
Cheers,
Tom
Apologies for the format of this post, was supposed to be a draft until I gathered a bit more info, but I accidentally sent the first message and was hastened. Anyway, there is a few things I'm looking into, firstly I'm using kernel26-ck with syslinux, and I also have intel_agp and i915 in mkinitcpio.conf for an earlier framebuffer. Using quiet seems to switch (or attempt to switch) to framebuffer earlier, my entire screen will go grey, then switch to black with a white box in the top left which looks to be 640x480 and then the boot messages will overwrite the white box and the system will commence to boot. Using loglevel=4, this does not occur. So there must be some difference somewhere, and although it's just initial framebuffer corruption it's fired up my curiosity. I've got a lot to try in order to pin this down, was just hoping somebody knew of any differences between quiet and loglevel=4 to save the fiddling. -- Jason Steadman http://www.meyithi.com/ http://twitter.com/meyithi
On Thu, Jul 28, 2011 at 11:12 AM, Meyithi <mail@meyithi.com> wrote:
On 28 July 2011 16:55, Tom Gundersen <teg@jklm.no> wrote:
Hi Jason,
On Thu, Jul 28, 2011 at 5:16 PM, Meyithi <mail@meyithi.com> wrote:
Does anybody know the difference between the quiet kernel parameter and loglevel=4 parameter?
All of the info I can find states that they should be identical, and /proc/sys/kernel/printk is identical.
They should be identical. Just look at the code: < http://git.kernel.org/?p=linux/kernel/git/torvalds/linux.git;a=blob;f=init/m...
. The quiet parameter just sets loglevel=4. Unless I'm missing something.
I am however getting framebuffer corruption during boot when using quiet which doesn't occur when using loglevel=4.
This is very odd. Any more info on this? Screenshots? Can you see a difference in dmesg?
Cheers,
Tom
Apologies for the format of this post, was supposed to be a draft until I gathered a bit more info, but I accidentally sent the first message and was hastened.
Anyway, there is a few things I'm looking into, firstly I'm using kernel26-ck with syslinux, and I also have intel_agp and i915 in mkinitcpio.conf for an earlier framebuffer. Using quiet seems to switch (or attempt to switch) to framebuffer earlier, my entire screen will go grey, then switch to black with a white box in the top left which looks to be 640x480 and then the boot messages will overwrite the white box and the system will commence to boot. Using loglevel=4, this does not occur. So there must be some difference somewhere, and although it's just initial framebuffer corruption it's fired up my curiosity.
I've got a lot to try in order to pin this down, was just hoping somebody knew of any differences between quiet and loglevel=4 to save the fiddling.
ah nice, i asked about this *exact* same thing back in May: http://mailman.archlinux.org/pipermail/arch-general/2011-May/020186.html ... unfortunately no one had an answer, and most said it wasn't happening to them, but you seem to have narrowed it down more maybe ... it happened on nvidia/ati/intel cards for me -- every single one of my systems in fact. i ended up just making a small mkinitcpio hook that ran first and cleared the buffer ... still a super brief period of corruption, but looked much better. the issue i had is identical to what you've described; interested to see what you find out. -- C Anthony
On 07/28/2011 08:13 PM, C Anthony Risinger wrote:
On Thu, Jul 28, 2011 at 11:12 AM, Meyithi<mail@meyithi.com> wrote:
On 28 July 2011 16:55, Tom Gundersen<teg@jklm.no> wrote:
Hi Jason,
On Thu, Jul 28, 2011 at 5:16 PM, Meyithi<mail@meyithi.com> wrote:
Does anybody know the difference between the quiet kernel parameter and loglevel=4 parameter?
All of the info I can find states that they should be identical, and /proc/sys/kernel/printk is identical.
They should be identical. Just look at the code: < http://git.kernel.org/?p=linux/kernel/git/torvalds/linux.git;a=blob;f=init/m...
. The quiet parameter just sets loglevel=4. Unless I'm missing something.
I am however getting framebuffer corruption during boot when using quiet which doesn't occur when using loglevel=4.
This is very odd. Any more info on this? Screenshots? Can you see a difference in dmesg?
Cheers,
Tom
Apologies for the format of this post, was supposed to be a draft until I gathered a bit more info, but I accidentally sent the first message and was hastened.
Anyway, there is a few things I'm looking into, firstly I'm using kernel26-ck with syslinux, and I also have intel_agp and i915 in mkinitcpio.conf for an earlier framebuffer. Using quiet seems to switch (or attempt to switch) to framebuffer earlier, my entire screen will go grey, then switch to black with a white box in the top left which looks to be 640x480 and then the boot messages will overwrite the white box and the system will commence to boot. Using loglevel=4, this does not occur. So there must be some difference somewhere, and although it's just initial framebuffer corruption it's fired up my curiosity.
I've got a lot to try in order to pin this down, was just hoping somebody knew of any differences between quiet and loglevel=4 to save the fiddling.
ah nice, i asked about this *exact* same thing back in May:
http://mailman.archlinux.org/pipermail/arch-general/2011-May/020186.html
... unfortunately no one had an answer, and most said it wasn't happening to them, but you seem to have narrowed it down more maybe ... it happened on nvidia/ati/intel cards for me -- every single one of my systems in fact.
i ended up just making a small mkinitcpio hook that ran first and cleared the buffer ... still a super brief period of corruption, but looked much better. the issue i had is identical to what you've described; interested to see what you find out.
I have this as well. 00:02.0 VGA compatible controller: Intel Corporation Mobile GME965/GLE960 Integrated Graphics Controller (rev 0c) Also with i915 and intel_agp in the ramdisk (for the same reason), also using 'quiet'. I won't reboot though, so no idea if 'loglevel=4' is different. This is not a new issue. I've seen it with the previous two or three kernels as well (2.6.37 - 2.6.39). I wasn't using 'quiet' before that, and had no such white box on the screen. Stock arch kernel, by the way, x86_64. -- cantabile "Jayne is a girl's name." -- River
On Thu, Jul 28, 2011 at 7:33 PM, cantabile <cantabile.desu@gmail.com> wrote:
Also with i915 and intel_agp in the ramdisk (for the same reason), also using 'quiet'.
It looks like I missed a site where the quiet parameter is parsed (looks very silly to do what they do if you ask me, should be looking at loglevel directly).
This is not a new issue. I've seen it with the previous two or three kernels as well (2.6.37 - 2.6.39). I wasn't using 'quiet' before that, and had no such white box on the screen.
Stock arch kernel, by the way, x86_64.
If the problem was not there in .36, but it was there in .37, I think the culprit must be 68f4d5a00adaab3. Could someone try reverting that on top of 3.0 and see if that fixes it? -t
On Thu, Jul 28, 2011 at 7:45 PM, Tom Gundersen <teg@jklm.no> wrote:
On Thu, Jul 28, 2011 at 7:33 PM, cantabile <cantabile.desu@gmail.com> wrote:
Also with i915 and intel_agp in the ramdisk (for the same reason), also using 'quiet'.
It looks like I missed a site where the quiet parameter is parsed (looks very silly to do what they do if you ask me, should be looking at loglevel directly).
This is not a new issue. I've seen it with the previous two or three kernels as well (2.6.37 - 2.6.39). I wasn't using 'quiet' before that, and had no such white box on the screen.
Stock arch kernel, by the way, x86_64.
If the problem was not there in .36, but it was there in .37, I think the culprit must be 68f4d5a00adaab3. Could someone try reverting that on top of 3.0 and see if that fixes it?
Nah, being silly, 8fee13a48e4879fba57725f6d9513df4bfa8e9f3 is more likely the culprit. Could anyone test? -t
This appears to be down to Syslinux, specifically a change introduced in 3.74 http://syslinux.zytor.com/wiki/index.php/Syslinux_3_Changelog#Changes_in_3.7... Suppress the Loading ... message if "quiet" is specified on the kernel
command line.
and the fix that we already knew is to use loglevel=4 instead of quiet http://syslinux.zytor.com/archives/2010-November/015688.html Use "loglevel=4" instead of "quiet"
It appears that the quiet kernel parameter also sets a quiet flag in Syslinux as well, which is doing something horrible to early framebuffers. As to the original question, it seems that "quiet" and "loglevel=4" are identical indeed, unless you use Syslinux. -- Jason Steadman http://www.meyithi.com/ http://twitter.com/meyithi
On Fri, Jul 29, 2011 at 3:43 AM, Meyithi <mail@meyithi.com> wrote:
It appears that the quiet kernel parameter also sets a quiet flag in Syslinux as well, which is doing something horrible to early framebuffers. As to the original question, it seems that "quiet" and "loglevel=4" are identical indeed, unless you use Syslinux.
nice catch -- that would make sense why all my machines behave the same way -- i *knew* syslinux was to blame!!! i'll try it out too, thanks. -- C Anthony
On Fri, Jul 29, 2011 at 10:43 AM, Meyithi <mail@meyithi.com> wrote:
It appears that the quiet kernel parameter also sets a quiet flag in Syslinux as well, which is doing something horrible to early framebuffers. As to the original question, it seems that "quiet" and "loglevel=4" are identical indeed, unless you use Syslinux.
For what it's worth: I'm using syslinux with quiet without seeing this problem (using i915). Also, the last kernel commit I pointed to in a previous email shows a place in the boot code where "quiet" is indeed different from "loglevel=4", just in case someone came across a different problem. Cheers, Tom
On Fri, Jul 29, 2011 at 3:35 PM, Tom Gundersen <teg@jklm.no> wrote:
On Fri, Jul 29, 2011 at 10:43 AM, Meyithi <mail@meyithi.com> wrote:
It appears that the quiet kernel parameter also sets a quiet flag in Syslinux as well, which is doing something horrible to early framebuffers. As to the original question, it seems that "quiet" and "loglevel=4" are identical indeed, unless you use Syslinux.
For what it's worth: I'm using syslinux with quiet without seeing this problem (using i915).
Also, the last kernel commit I pointed to in a previous email shows a place in the boot code where "quiet" is indeed different from "loglevel=4", just in case someone came across a different problem.
i finally got around to trying this; it eliminated the "white box" effect for me, on these two machines anyway (nvidia/radeon) thanks! ... i'm slightly curious why you *don't* see it Tom :-), i haven't made the change on our netbook yet, which also uses i915, so we'll see i suppose. -- C Anthony
On Sat, Aug 6, 2011 at 6:07 AM, C Anthony Risinger <anthony@xtfx.me> wrote:
On Fri, Jul 29, 2011 at 3:35 PM, Tom Gundersen <teg@jklm.no> wrote:
On Fri, Jul 29, 2011 at 10:43 AM, Meyithi <mail@meyithi.com> wrote:
It appears that the quiet kernel parameter also sets a quiet flag in Syslinux as well, which is doing something horrible to early framebuffers. As to the original question, it seems that "quiet" and "loglevel=4" are identical indeed, unless you use Syslinux.
For what it's worth: I'm using syslinux with quiet without seeing this problem (using i915).
Also, the last kernel commit I pointed to in a previous email shows a place in the boot code where "quiet" is indeed different from "loglevel=4", just in case someone came across a different problem.
i finally got around to trying this; it eliminated the "white box" effect for me, on these two machines anyway (nvidia/radeon)
Sorry, what did you try? Revert the commit? -t
On Sat, Aug 6, 2011 at 7:09 AM, Tom Gundersen <teg@jklm.no> wrote:
On Sat, Aug 6, 2011 at 6:07 AM, C Anthony Risinger <anthony@xtfx.me> wrote:
On Fri, Jul 29, 2011 at 3:35 PM, Tom Gundersen <teg@jklm.no> wrote:
On Fri, Jul 29, 2011 at 10:43 AM, Meyithi <mail@meyithi.com> wrote:
It appears that the quiet kernel parameter also sets a quiet flag in Syslinux as well, which is doing something horrible to early framebuffers. As to the original question, it seems that "quiet" and "loglevel=4" are identical indeed, unless you use Syslinux.
For what it's worth: I'm using syslinux with quiet without seeing this problem (using i915).
Also, the last kernel commit I pointed to in a previous email shows a place in the boot code where "quiet" is indeed different from "loglevel=4", just in case someone came across a different problem.
i finally got around to trying this; it eliminated the "white box" effect for me, on these two machines anyway (nvidia/radeon)
Sorry, what did you try? Revert the commit?
beh sorry about that -- completely wrong context no i only removed `quiet` and added `loglevel=4` ... this alone eliminated the artifacts. -- C Anthony
participants (4)
-
C Anthony Risinger
-
cantabile
-
Meyithi
-
Tom Gundersen