[PATCH 0/16] Assorted patches

[PATCH 0/16] Assorted patches

[Jan Engelhardt]

Hello list,


People collect stamps, etc. and I do patches. It is April 01, but let ye 
know that even the patch that would get the most naks is not really 
meant as an April joke.
Most of them are independent, but I'll let `quilt graph` judge their 
dependencies.

	[01/] vt-sysfs-for-colors.diff
	[02/] vt-pure-colors.diff
	[03/] vt-underline-color.diff
	[04/] vt-printk-color.diff
	[05/] fix-kthread-niceness.diff
	[06/] isofs-add-write-bit.diff
	[07/] kconfig-dynamic-frequency.diff
	[08/] console-printk-level.diff
	[09/] zlib-decompression-status.diff
	[10/] show-partitions-on-mount-error.diff
	[11/] samba-eintr-fix.diff
	[12/] cifs-use-mutex.diff
	[13/] show-pipesize-in-stat.diff
	[14/] kconfig-allow-override.diff
	[15/] use-regular-eth-suffix.diff
	[16/] warn-on-kthread-name-truncation.diff

(`quilt graph` says the order of the vt-* ones is important.)


Jan
-- 

[PATCH 1/16] vt-sysfs-for-colors.diff

[Jan Engelhardt]

Allow for the palette to be exposed and changed via sysfs. A call to
/usr/bin/reset will slurp the new definitions in for the current
console.

Already posted at http://lkml.org/lkml/2006/1/15/149

Signed-off-by: Jan Engelhardt <jengelh@gmx.de>

 vt.c |    4 ++++
 1 file changed, 4 insertions(+)

Index: linux-2.6.21-rc5/drivers/char/vt.c
===================================================================
--- linux-2.6.21-rc5.orig/drivers/char/vt.c
+++ linux-2.6.21-rc5/drivers/char/vt.c
@@ -934,6 +934,10 @@ int default_grn[] = {0x00,0x00,0xaa,0x55
 int default_blu[] = {0x00,0x00,0x00,0x00,0xaa,0xaa,0xaa,0xaa,
     0x55,0x55,0x55,0x55,0xff,0xff,0xff,0xff};
 
+module_param_array(default_red, int, NULL, S_IRUGO | S_IWUSR);
+module_param_array(default_grn, int, NULL, S_IRUGO | S_IWUSR);
+module_param_array(default_blu, int, NULL, S_IRUGO | S_IWUSR);
+
 /*
  * gotoxy() must verify all boundaries, because the arguments
  * might also be negative. If the given position is out of
#<EOF>

Re: [PATCH 1/16] vt-sysfs-for-colors.diff

[Antonino A. Daplas]

On Sun, 2007-04-01 at 20:13 +0200, Jan Engelhardt wrote:
> Allow for the palette to be exposed and changed via sysfs. A call to
> /usr/bin/reset will slurp the new definitions in for the current
> console.
> 

I like this. The escape sequences to change the palette does not stay
permanently.

Tony

[PATCH 17/16] Do not reset UTF8 on terminal reset

[Jan Engelhardt]

On Apr 2 2007 14:49, Antonino A. Daplas wrote:
>On Sun, 2007-04-01 at 20:13 +0200, Jan Engelhardt wrote:
>
>> Allow for the palette to be exposed and changed via sysfs. A call to
>> /usr/bin/reset will slurp the new definitions in for the current
>> console.
>
>I like this. The escape sequences to change the palette does not stay
>permanently.

As much as you like it, there is a slight problem with it. Linux will
currently throw virtual consoles out of UTF8 mode when reset is called.

Also see:
	http://lkml.org/lkml/2005/5/17/289
	http://lkml.org/lkml/2005/5/17/297

These posts argue about terminfo being the culprit. But how can terminfo 
be at fault, when `echo -en "\ec"` triggers it too?

Since I am in a patch mood, here's my stance/patch, which is compile and 
run tested and behaves as expected (both with `echo` and `reset`).

Signed-off-by: Jan Engelhardt <jengelh@gmx.de>
Cc: Andrew Morton <akpm@osdl.org>
Cc: Daniel Jacobowitz <dan@debian.org>
Cc: Paul LeoNerd Evans <leonerd@leonerd.org.uk>
Obsoletes: https://bugzilla.novell.com/show_bug.cgi?id=225853

Index: linux-2.6.21-rc5/drivers/char/vt.c
===================================================================
--- linux-2.6.21-rc5.orig/drivers/char/vt.c
+++ linux-2.6.21-rc5/drivers/char/vt.c
@@ -1518,7 +1518,6 @@ static void reset_terminal(struct vc_dat
 	vc->vc_charset		= 0;
 	vc->vc_need_wrap	= 0;
 	vc->vc_report_mouse	= 0;
-	vc->vc_utf		= 0;
 	vc->vc_utf_count	= 0;
 
 	vc->vc_disp_ctrl	= 0;
#<EOF>

Re: [PATCH 17/16] Do not reset UTF8 on terminal reset

[Antonino A. Daplas]

On Mon, 2007-04-02 at 09:31 +0200, Jan Engelhardt wrote:
> On Apr 2 2007 14:49, Antonino A. Daplas wrote:
> >On Sun, 2007-04-01 at 20:13 +0200, Jan Engelhardt wrote:
> >
> >> Allow for the palette to be exposed and changed via sysfs. A call to
> >> /usr/bin/reset will slurp the new definitions in for the current
> >> console.
> >
> >I like this. The escape sequences to change the palette does not stay
> >permanently.
> 
> As much as you like it, there is a slight problem with it. Linux will
> currently throw virtual consoles out of UTF8 mode when reset is called.
> 
> Also see:
> 	http://lkml.org/lkml/2005/5/17/289
> 	http://lkml.org/lkml/2005/5/17/297
> 
> These posts argue about terminfo being the culprit. But how can terminfo 
> be at fault, when `echo -en "\ec"` triggers it too?
> 
> Since I am in a patch mood, here's my stance/patch, which is compile and 
> run tested and behaves as expected (both with `echo` and `reset`).

Tested and works here too.

Tony

Re: [PATCH 17/16] Do not reset UTF8 on terminal reset

[Paul LeoNerd Evans]

[-- Attachment #1: Type: text/plain, Size: 1352 bytes --]

On Mon, Apr 02, 2007 at 09:31:37AM +0200, Jan Engelhardt wrote:
> Also see:
> 	http://lkml.org/lkml/2005/5/17/289
> 	http://lkml.org/lkml/2005/5/17/297
> 
> These posts argue about terminfo being the culprit. But how can terminfo 
> be at fault, when `echo -en "\ec"` triggers it too?
> 
> Since I am in a patch mood, here's my stance/patch, which is compile and 
> run tested and behaves as expected (both with `echo` and `reset`).

...

> Index: linux-2.6.21-rc5/drivers/char/vt.c
> ===================================================================
> --- linux-2.6.21-rc5.orig/drivers/char/vt.c
> +++ linux-2.6.21-rc5/drivers/char/vt.c
> @@ -1518,7 +1518,6 @@ static void reset_terminal(struct vc_dat
>  	vc->vc_charset		= 0;
>  	vc->vc_need_wrap	= 0;
>  	vc->vc_report_mouse	= 0;
> -	vc->vc_utf		= 0;
>  	vc->vc_utf_count	= 0;
>  
>  	vc->vc_disp_ctrl	= 0;
> #<EOF>

Is it OK to do that? I recall when I was originally looking at the code
I didn't want to just remove that line, because it looked like that was
being used to first initialise the vc* structure when it is created, as
well as reset it every time. Doesn't this leave vc->vc_utf uninitialised
when a new VC is allocated?

-- 
Paul "LeoNerd" Evans

leonerd@leonerd.org.uk
ICQ# 4135350       |  Registered Linux# 179460
http://www.leonerd.org.uk/

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [PATCH 17/16] Do not reset UTF8 on terminal reset

[Antonino A. Daplas]

[-- Attachment #1: Type: text/plain, Size: 1396 bytes --]

On Mon, 2007-04-02 at 12:26 +0100, Paul LeoNerd Evans wrote:
> On Mon, Apr 02, 2007 at 09:31:37AM +0200, Jan Engelhardt wrote:
> > Also see:
> > 	http://lkml.org/lkml/2005/5/17/289
> > 	http://lkml.org/lkml/2005/5/17/297
> > 
> > These posts argue about terminfo being the culprit. But how can terminfo 
> > be at fault, when `echo -en "\ec"` triggers it too?
> > 
> > Since I am in a patch mood, here's my stance/patch, which is compile and 
> > run tested and behaves as expected (both with `echo` and `reset`).
> 
> ...
> 
> > Index: linux-2.6.21-rc5/drivers/char/vt.c
> > ===================================================================
> > --- linux-2.6.21-rc5.orig/drivers/char/vt.c
> > +++ linux-2.6.21-rc5/drivers/char/vt.c
> > @@ -1518,7 +1518,6 @@ static void reset_terminal(struct vc_dat
> >  	vc->vc_charset		= 0;
> >  	vc->vc_need_wrap	= 0;
> >  	vc->vc_report_mouse	= 0;
> > -	vc->vc_utf		= 0;
> >  	vc->vc_utf_count	= 0;
> >  
> >  	vc->vc_disp_ctrl	= 0;
> > #<EOF>
> 
> Is it OK to do that? I recall when I was originally looking at the code
> I didn't want to just remove that line, because it looked like that was
> being used to first initialise the vc* structure when it is created, as
> well as reset it every time. Doesn't this leave vc->vc_utf uninitialised
> when a new VC is allocated?

That's true.  We can move the line vc->vc_utf = 0; in vc_init()
instead. 

Tony


[-- Attachment #2: vc_reset.diff --]
[-- Type: text/x-patch, Size: 576 bytes --]

diff --git a/drivers/char/vt.c b/drivers/char/vt.c
index 1bbb45b..23e4432 100644
--- a/drivers/char/vt.c
+++ b/drivers/char/vt.c
@@ -1497,7 +1497,6 @@ static void reset_terminal(struct vc_dat
 	vc->vc_charset		= 0;
 	vc->vc_need_wrap	= 0;
 	vc->vc_report_mouse	= 0;
-	vc->vc_utf		= 0;
 	vc->vc_utf_count	= 0;
 
 	vc->vc_disp_ctrl	= 0;
@@ -2590,6 +2589,7 @@ static void vc_init(struct vc_data *vc, 
 	vc->vc_rows = rows;
 	vc->vc_size_row = cols << 1;
 	vc->vc_screenbuf_size = vc->vc_rows * vc->vc_size_row;
+	vc->vc_utf = 0;
 
 	set_origin(vc);
 	vc->vc_pos = vc->vc_origin;

Re: [PATCH 17/16] Do not reset UTF8 on terminal reset

[Paul LeoNerd Evans]

[-- Attachment #1: Type: text/plain, Size: 1931 bytes --]

On Mon, Apr 02, 2007 at 07:44:13PM +0800, Antonino A. Daplas wrote:
> > Is it OK to do that? I recall when I was originally looking at the code
> > I didn't want to just remove that line, because it looked like that was
> > being used to first initialise the vc* structure when it is created, as
> > well as reset it every time. Doesn't this leave vc->vc_utf uninitialised
> > when a new VC is allocated?
> 
> That's true.  We can move the line vc->vc_utf = 0; in vc_init()
> instead. 

...

> @@ -2590,6 +2589,7 @@ static void vc_init(struct vc_data *vc, 
>  	vc->vc_rows = rows;
>  	vc->vc_size_row = cols << 1;
>  	vc->vc_screenbuf_size = vc->vc_rows * vc->vc_size_row;
> +	vc->vc_utf = 0;
>  
>  	set_origin(vc);
>  	vc->vc_pos = vc->vc_origin;

While we're on that subject, did you take a look at my original mail?
The intent with that patch was to allow system policy to state all new
VCs are UTF-8-enabled by default. I feel that in 2007 this should be the
default setting.

Would it therefore be possible to have

   vc->vc_utf = some_default;

where some_default comes maybe from a sysctl or some other configurable
source? Or maybe even have a compiletime option?

This would get around many bugs. For example, on boot, "unicode_start"
can only set utf8 mode on the existing VCs 1 to 6. If X11 fails to
start, debian nicely runs me the XKeepsCrashing program, which offers to
show me logs and the like. It reads the locale, en_GB.UTF-8 and
determines we're in UTF-8 mode, so outputs Unicode linedrawing
characters for dialogs. Unfortunately, we're on VC7 which doesn't have
UTF-8 mode turned on, so much mess results.

It would be nice if the kernel's default UTF-8 mode for new VCs could be
synched to whatever local policy was regarding locale.

-- 
Paul "LeoNerd" Evans

leonerd@leonerd.org.uk
ICQ# 4135350       |  Registered Linux# 179460
http://www.leonerd.org.uk/

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [PATCH 17/16] Do not reset UTF8 on terminal reset

[Antonino A. Daplas]

[-- Attachment #1: Type: text/plain, Size: 2160 bytes --]

On Mon, 2007-04-02 at 12:54 +0100, Paul LeoNerd Evans wrote:
> On Mon, Apr 02, 2007 at 07:44:13PM +0800, Antonino A. Daplas wrote:
> > > Is it OK to do that? I recall when I was originally looking at the code
> > > I didn't want to just remove that line, because it looked like that was
> > > being used to first initialise the vc* structure when it is created, as
> > > well as reset it every time. Doesn't this leave vc->vc_utf uninitialised
> > > when a new VC is allocated?
> > 
> > That's true.  We can move the line vc->vc_utf = 0; in vc_init()
> > instead. 
> 
> ...
> 
> > @@ -2590,6 +2589,7 @@ static void vc_init(struct vc_data *vc, 
> >  	vc->vc_rows = rows;
> >  	vc->vc_size_row = cols << 1;
> >  	vc->vc_screenbuf_size = vc->vc_rows * vc->vc_size_row;
> > +	vc->vc_utf = 0;
> >  
> >  	set_origin(vc);
> >  	vc->vc_pos = vc->vc_origin;
> 
> While we're on that subject, did you take a look at my original mail?

I did not actually.

> The intent with that patch was to allow system policy to state all new
> VCs are UTF-8-enabled by default. I feel that in 2007 this should be the
> default setting.
> 
> Would it therefore be possible to have
> 
>    vc->vc_utf = some_default;
> 
> where some_default comes maybe from a sysctl or some other configurable
> source? Or maybe even have a compiletime option?
> 
> This would get around many bugs. For example, on boot, "unicode_start"
> can only set utf8 mode on the existing VCs 1 to 6. If X11 fails to
> start, debian nicely runs me the XKeepsCrashing program, which offers to
> show me logs and the like. It reads the locale, en_GB.UTF-8 and
> determines we're in UTF-8 mode, so outputs Unicode linedrawing
> characters for dialogs. Unfortunately, we're on VC7 which doesn't have
> UTF-8 mode turned on, so much mess results.
> 
> It would be nice if the kernel's default UTF-8 mode for new VCs could be
> synched to whatever local policy was regarding locale.

How about this? This should allow us to reset the terminal without
affecting utf, and also allows setting of a global utf default that can
be set via boot option or sysfs:

echo 1 > /sys/module/vt/parameters/default_utf

Tony




> 

[-- Attachment #2: utf.diff --]
[-- Type: text/x-patch, Size: 913 bytes --]

diff --git a/drivers/char/vt.c b/drivers/char/vt.c
index 1bbb45b..b9b3247 100644
--- a/drivers/char/vt.c
+++ b/drivers/char/vt.c
@@ -157,6 +157,8 @@ static void blank_screen_t(unsigned long
 static void set_palette(struct vc_data *vc);
 
 static int printable;		/* Is console ready for printing? */
+static int default_utf;
+module_param(default_utf, int, S_IRUGO | S_IWUSR);
 
 /*
  * ignore_poke: don't unblank the screen when things are typed.  This is
@@ -1497,7 +1499,6 @@ static void reset_terminal(struct vc_dat
 	vc->vc_charset		= 0;
 	vc->vc_need_wrap	= 0;
 	vc->vc_report_mouse	= 0;
-	vc->vc_utf		= 0;
 	vc->vc_utf_count	= 0;
 
 	vc->vc_disp_ctrl	= 0;
@@ -2590,6 +2591,7 @@ static void vc_init(struct vc_data *vc, 
 	vc->vc_rows = rows;
 	vc->vc_size_row = cols << 1;
 	vc->vc_screenbuf_size = vc->vc_rows * vc->vc_size_row;
+	vc->vc_utf = default_utf;
 
 	set_origin(vc);
 	vc->vc_pos = vc->vc_origin;

Re: [PATCH 17/16] Do not reset UTF8 on terminal reset

[Paul LeoNerd Evans]

[-- Attachment #1: Type: text/plain, Size: 579 bytes --]

On Mon, Apr 02, 2007 at 08:47:59PM +0800, Antonino A. Daplas wrote:
> How about this? This should allow us to reset the terminal without
> affecting utf, and also allows setting of a global utf default that can
> be set via boot option or sysfs:
> 
> echo 1 > /sys/module/vt/parameters/default_utf

That's almost exactly what my suggestion was :)

(except that mine went into /proc/tty/vt/default_utf8_mode, but anywhere
is good really)

-- 
Paul "LeoNerd" Evans

leonerd@leonerd.org.uk
ICQ# 4135350       |  Registered Linux# 179460
http://www.leonerd.org.uk/

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [PATCH 17/16] Do not reset UTF8 on terminal reset

[Jan Engelhardt]

On Apr 2 2007 20:47, Antonino A. Daplas wrote:
>> 
>> It would be nice if the kernel's default UTF-8 mode for new VCs could be
>> synched to whatever local policy was regarding locale.
>
>How about this? This should allow us to reset the terminal without
>affecting utf, and also allows setting of a global utf default that can
>be set via boot option or sysfs:
>
>echo 1 > /sys/module/vt/parameters/default_utf

Might want to call it "default_utf8", but it's just minor.

Tony, if you have a patch, send it along so I can test it under "special local
circumstances" :-)

Jan
-- 

Re: [PATCH 17/16] Do not reset UTF8 on terminal reset

[Pavel Machek]

Hi!

> >> Allow for the palette to be exposed and changed via sysfs. A call to
> >> /usr/bin/reset will slurp the new definitions in for the current
> >> console.
> >
> >I like this. The escape sequences to change the palette does not stay
> >permanently.
> 
> As much as you like it, there is a slight problem with it. Linux will
> currently throw virtual consoles out of UTF8 mode when reset is called.
> 
> Also see:
> 	http://lkml.org/lkml/2005/5/17/289
> 	http://lkml.org/lkml/2005/5/17/297
> 
> These posts argue about terminfo being the culprit. But how can terminfo 
> be at fault, when `echo -en "\ec"` triggers it too?
> 
> Since I am in a patch mood, here's my stance/patch, which is compile and 
> run tested and behaves as expected (both with `echo` and `reset`).

Should not reset actually reset terminal? This introduces
bug/unexpected behaviour...
							Pavel
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

[PATCH 2/16] vt-pure-colors.diff

[Jan Engelhardt]

Have the Linux kernel set a new VGA palette for the first 16 colors.
The new values reduce the saturation (white component) and therefore
increase contrast.

Already posted once: http://lkml.org/lkml/2006/1/15/149

Signed-off-by: Jan Engelhardt <jengelh@gmx.de>

 vt.c |   12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)

Index: linux-2.6.21-rc5/drivers/char/vt.c
===================================================================
--- linux-2.6.21-rc5.orig/drivers/char/vt.c
+++ linux-2.6.21-rc5/drivers/char/vt.c
@@ -927,12 +927,12 @@ unsigned char color_table[] = { 0, 4, 2,
 				       8,12,10,14, 9,13,11,15 };
 
 /* the default colour table, for VGA+ colour systems */
-int default_red[] = {0x00,0xaa,0x00,0xaa,0x00,0xaa,0x00,0xaa,
-    0x55,0xff,0x55,0xff,0x55,0xff,0x55,0xff};
-int default_grn[] = {0x00,0x00,0xaa,0x55,0x00,0x00,0xaa,0xaa,
-    0x55,0x55,0xff,0xff,0x55,0x55,0xff,0xff};
-int default_blu[] = {0x00,0x00,0x00,0x00,0xaa,0xaa,0xaa,0xaa,
-    0x55,0x55,0x55,0x55,0xff,0xff,0xff,0xff};
+int default_red[] =
+{0x00,0xaa,0x00,0xaa,0x00,0xaa,0x00,0xaa,0x55,0xff,0x00,0xff,0x00,0xff,0x00,0xff};
+int default_grn[] =
+{0x00,0x00,0xaa,0x55,0x00,0x00,0xaa,0xaa,0x55,0x00,0xff,0xff,0x00,0x00,0xff,0xff};
+int default_blu[] =
+{0x00,0x00,0x00,0x00,0x90,0xaa,0xaa,0xaa,0x55,0x00,0x00,0x00,0xff,0xff,0xff,0xff};
 
 module_param_array(default_red, int, NULL, S_IRUGO | S_IWUSR);
 module_param_array(default_grn, int, NULL, S_IRUGO | S_IWUSR);
#<EOF>

Re: [PATCH 2/16] vt-pure-colors.diff

[James Bruce]

Jan Engelhardt wrote:
> Have the Linux kernel set a new VGA palette for the first 16 colors.
> The new values reduce the saturation (white component) and therefore
> increase contrast.
<snip>

While the patch seems fine, this comment is not correct.  The patch is 
decreasing the *brightness* in order to *increase* saturation. 
Saturation[1] is the deepness of a color, not the white component 
(brightness, lightness).

  - Jim Bruce

[1] http://www.poynton.com/notes/colour_and_gamma/ColorFAQ.html#RTFToC6

Re: [PATCH 2/16] vt-pure-colors.diff

[Antonino A. Daplas]

On Sun, 2007-04-01 at 20:14 +0200, Jan Engelhardt wrote:
> Have the Linux kernel set a new VGA palette for the first 16 colors.
> The new values reduce the saturation (white component) and therefore
> increase contrast.
> 
> Already posted once: http://lkml.org/lkml/2006/1/15/149
> 
> Signed-off-by: Jan Engelhardt <jengelh@gmx.de>
> 
>  vt.c |   12 ++++++------
>  1 file changed, 6 insertions(+), 6 deletions(-)
> 
> Index: linux-2.6.21-rc5/drivers/char/vt.c
> ===================================================================
> --- linux-2.6.21-rc5.orig/drivers/char/vt.c
> +++ linux-2.6.21-rc5/drivers/char/vt.c
> @@ -927,12 +927,12 @@ unsigned char color_table[] = { 0, 4, 2,
>  				       8,12,10,14, 9,13,11,15 };
>  
>  /* the default colour table, for VGA+ colour systems */
> -int default_red[] = {0x00,0xaa,0x00,0xaa,0x00,0xaa,0x00,0xaa,
> -    0x55,0xff,0x55,0xff,0x55,0xff,0x55,0xff};
> -int default_grn[] = {0x00,0x00,0xaa,0x55,0x00,0x00,0xaa,0xaa,
> -    0x55,0x55,0xff,0xff,0x55,0x55,0xff,0xff};
> -int default_blu[] = {0x00,0x00,0x00,0x00,0xaa,0xaa,0xaa,0xaa,
> -    0x55,0x55,0x55,0x55,0xff,0xff,0xff,0xff};
> +int default_red[] =
> +{0x00,0xaa,0x00,0xaa,0x00,0xaa,0x00,0xaa,0x55,0xff,0x00,0xff,0x00,0xff,0x00,0xff};
> +int default_grn[] =
> +{0x00,0x00,0xaa,0x55,0x00,0x00,0xaa,0xaa,0x55,0x00,0xff,0xff,0x00,0x00,0xff,0xff};
> +int default_blu[] =
> +{0x00,0x00,0x00,0x00,0x90,0xaa,0xaa,0xaa,0x55,0x00,0x00,0x00,0xff,0xff,0xff,0xff};
>  

Not too sure about this, the default palette is based on ANSI/ECMA
standards (at least for the first 8 colors). Also, this will have subtle
effects on the 16-color logo. Since your first patch already allows us
to change the palette, let's leave the default alone.

 
Tony

Re: [PATCH 2/16] vt-pure-colors.diff

[Jan Engelhardt]

On Apr 2 2007 14:49, Antonino A. Daplas wrote:
>
>Not too sure about this, the default palette is based on ANSI/ECMA
>standards (at least for the first 8 colors). Also, this will have subtle
>effects on the 16-color logo. Since your first patch already allows us
>to change the palette, let's leave the default alone.

Ok.
But to note: xterm and anything graphical (including MS Windows)
uses the anti-pale color (255,0,0) instead of (255,170,170).


Jan
-- 

Re: [PATCH 2/16] vt-pure-colors.diff

[Antonino A. Daplas]

On Mon, 2007-04-02 at 09:04 +0200, Jan Engelhardt wrote:
> On Apr 2 2007 14:49, Antonino A. Daplas wrote:
> >
> >Not too sure about this, the default palette is based on ANSI/ECMA
> >standards (at least for the first 8 colors). Also, this will have subtle
> >effects on the 16-color logo. Since your first patch already allows us
> >to change the palette, let's leave the default alone.
> 
> Ok.
> But to note: xterm and anything graphical (including MS Windows)
> uses the anti-pale color (255,0,0) instead of (255,170,170).

Actually, what threw me off was index 4 in default_blu, which has a
value of 0x90.  This looks weird to me, but I guess that is a typo. I
really have nothing against using 0, 0, 255 instead of 170, 170, 255 in
the intense version of the first 8.

Tony 

Re: [PATCH 2/16] vt-pure-colors.diff

[Jan Engelhardt]

On Apr 2 2007 15:50, Antonino A. Daplas wrote:
>On Mon, 2007-04-02 at 09:04 +0200, Jan Engelhardt wrote:
>> On Apr 2 2007 14:49, Antonino A. Daplas wrote:
>> >
>> >Not too sure about this, the default palette is based on ANSI/ECMA
>> >standards (at least for the first 8 colors). Also, this will have subtle
>> >effects on the 16-color logo. Since your first patch already allows us
>> >to change the palette, let's leave the default alone.
>> 
>> Ok.
>> But to note: xterm and anything graphical (including MS Windows)
>> uses the anti-pale color (255,0,0) instead of (255,170,170).
>
>Actually, what threw me off was index 4 in default_blu, which has a
>value of 0x90.  This looks weird to me, but I guess that is a typo. I
>really have nothing against using 0, 0, 255 instead of 170, 170, 255 in
>the intense version of the first 8.

That one is a psychological thing. :-) People like blue backgrounds, because a
green or red one would not really fit as a bg color for Midnight Commander,
pine and your-favorite-editor. So you choose blue because it fits nicely. (See
http://img81.imageshack.us/img81/8858/blueqj6.png for an example, and compare
with http://img81.imageshack.us/img81/7498/reddm1.png and
http://img142.imageshack.us/img142/8470/greennq2.png for counterexamples. Even
a bright blue (0,0,192) is still better than red/green, the reason being that
blue is much less percepted by the eye than green or red. You will find various
formulas on the 'net that will convert from color to grayscale - these formulas
have the 'perception' levels, which are, summarized, 76*r, 151*g, 26*b on a 255
scale.) Personally I find the regular blue (0,0,170/0xA0) a little too bright
so I turned it down to 0x90. Adjust for your preference.


Jan
-- 

[PATCH 3/16] vt-underline-color.diff

[Jan Engelhardt]

Add color support to the "underline" and "italic" attributes as in
OpenBSD/NetBSD-style (vt220) and xterm.

Signed-off-by: Jan Engelhardt <jengelh@gmx.de>

 drivers/char/vt.c               |   34 ++++++++++++++++++++++++++++------
 drivers/video/console/mdacon.c  |    3 ++-
 drivers/video/console/promcon.c |    3 ++-
 drivers/video/console/sticon.c  |    2 +-
 drivers/video/console/vgacon.c  |   12 ++++++++----
 include/linux/console.h         |    2 +-
 include/linux/console_struct.h  |    3 +++
 7 files changed, 45 insertions(+), 14 deletions(-)

Index: linux-2.6.21-rc5/drivers/char/vt.c
===================================================================
--- linux-2.6.21-rc5.orig/drivers/char/vt.c
+++ linux-2.6.21-rc5/drivers/char/vt.c
@@ -348,10 +348,12 @@ void update_region(struct vc_data *vc, u
 
 /* Structure of attributes is hardware-dependent */
 
-static u8 build_attr(struct vc_data *vc, u8 _color, u8 _intensity, u8 _blink, u8 _underline, u8 _reverse)
+static u8 build_attr(struct vc_data *vc, u8 _color, u8 _intensity, u8 _blink,
+    u8 _underline, u8 _reverse, u8 _italic)
 {
 	if (vc->vc_sw->con_build_attr)
-		return vc->vc_sw->con_build_attr(vc, _color, _intensity, _blink, _underline, _reverse);
+		return vc->vc_sw->con_build_attr(vc, _color, _intensity,
+		       _blink, _underline, _reverse, _italic);
 
 #ifndef VT_BUF_VRAM_ONLY
 /*
@@ -368,10 +370,13 @@ static u8 build_attr(struct vc_data *vc,
 	u8 a = vc->vc_color;
 	if (!vc->vc_can_do_color)
 		return _intensity |
+		       (_italic ? 2 : 0) |
 		       (_underline ? 4 : 0) |
 		       (_reverse ? 8 : 0) |
 		       (_blink ? 0x80 : 0);
-	if (_underline)
+	if (_italic)
+		a = (a & 0xF0) | vc->vc_itcolor;
+	else if (_underline)
 		a = (a & 0xf0) | vc->vc_ulcolor;
 	else if (_intensity == 0)
 		a = (a & 0xf0) | vc->vc_ulcolor;
@@ -392,8 +397,10 @@ static u8 build_attr(struct vc_data *vc,
 
 static void update_attr(struct vc_data *vc)
 {
-	vc->vc_attr = build_attr(vc, vc->vc_color, vc->vc_intensity, vc->vc_blink, vc->vc_underline, vc->vc_reverse ^ vc->vc_decscnm);
-	vc->vc_video_erase_char = (build_attr(vc, vc->vc_color, 1, vc->vc_blink, 0, vc->vc_decscnm) << 8) | ' ';
+	vc->vc_attr = build_attr(vc, vc->vc_color, vc->vc_intensity,
+	              vc->vc_blink, vc->vc_underline,
+	              vc->vc_reverse ^ vc->vc_decscnm, vc->vc_italic);
+	vc->vc_video_erase_char = (build_attr(vc, vc->vc_color, 1, vc->vc_blink, 0, vc->vc_decscnm, 0) << 8) | ' ';
 }
 
 /* Note: inverting the screen twice should revert to the original state */
@@ -1136,6 +1143,7 @@ static void csi_X(struct vc_data *vc, in
 static void default_attr(struct vc_data *vc)
 {
 	vc->vc_intensity = 1;
+	vc->vc_italic = 0;
 	vc->vc_underline = 0;
 	vc->vc_reverse = 0;
 	vc->vc_blink = 0;
@@ -1158,6 +1166,9 @@ static void csi_m(struct vc_data *vc)
 			case 2:
 				vc->vc_intensity = 0;
 				break;
+			case 3:
+				vc->vc_italic = 1;
+				break;
 			case 4:
 				vc->vc_underline = 1;
 				break;
@@ -1198,6 +1209,9 @@ static void csi_m(struct vc_data *vc)
 			case 22:
 				vc->vc_intensity = 1;
 				break;
+			case 23:
+				vc->vc_italic = 0;
+				break;
 			case 24:
 				vc->vc_underline = 0;
 				break;
@@ -1458,6 +1472,7 @@ static void save_cur(struct vc_data *vc)
 	vc->vc_saved_x		= vc->vc_x;
 	vc->vc_saved_y		= vc->vc_y;
 	vc->vc_s_intensity	= vc->vc_intensity;
+	vc->vc_s_italic         = vc->vc_italic;
 	vc->vc_s_underline	= vc->vc_underline;
 	vc->vc_s_blink		= vc->vc_blink;
 	vc->vc_s_reverse	= vc->vc_reverse;
@@ -1472,6 +1487,7 @@ static void restore_cur(struct vc_data *
 {
 	gotoxy(vc, vc->vc_saved_x, vc->vc_saved_y);
 	vc->vc_intensity	= vc->vc_s_intensity;
+	vc->vc_italic		= vc->vc_s_italic;
 	vc->vc_underline	= vc->vc_s_underline;
 	vc->vc_blink		= vc->vc_s_blink;
 	vc->vc_reverse		= vc->vc_s_reverse;
@@ -2585,6 +2601,11 @@ static void con_close(struct tty_struct 
 	mutex_unlock(&tty_mutex);
 }
 
+static int default_italic_color    = 2; // green (ASCII)
+static int default_underline_color = 3; // cyan (ASCII)
+module_param_named(italic, default_italic_color, int, S_IRUGO | S_IWUSR);
+module_param_named(underline, default_underline_color, int, S_IRUGO | S_IWUSR);
+
 static void vc_init(struct vc_data *vc, unsigned int rows,
 		    unsigned int cols, int do_clear)
 {
@@ -2604,7 +2625,8 @@ static void vc_init(struct vc_data *vc, 
 		vc->vc_palette[k++] = default_blu[j] ;
 	}
 	vc->vc_def_color       = 0x07;   /* white */
-	vc->vc_ulcolor		= 0x0f;   /* bold white */
+	vc->vc_ulcolor         = default_underline_color;
+	vc->vc_itcolor         = default_italic_color;
 	vc->vc_halfcolor       = 0x08;   /* grey */
 	init_waitqueue_head(&vc->paste_wait);
 	reset_terminal(vc, do_clear);
Index: linux-2.6.21-rc5/drivers/video/console/mdacon.c
===================================================================
--- linux-2.6.21-rc5.orig/drivers/video/console/mdacon.c
+++ linux-2.6.21-rc5/drivers/video/console/mdacon.c
@@ -384,7 +384,7 @@ static inline u16 mda_convert_attr(u16 c
 }
 
 static u8 mdacon_build_attr(struct vc_data *c, u8 color, u8 intensity, 
-			    u8 blink, u8 underline, u8 reverse)
+			    u8 blink, u8 underline, u8 reverse, u8 italic)
 {
 	/* The attribute is just a bit vector:
 	 *
@@ -397,6 +397,7 @@ static u8 mdacon_build_attr(struct vc_da
 	return (intensity & 3) |
 		((underline & 1) << 2) |
 		((reverse   & 1) << 3) |
+		(!!italic << 4) |
 		((blink     & 1) << 7);
 }
 
Index: linux-2.6.21-rc5/drivers/video/console/promcon.c
===================================================================
--- linux-2.6.21-rc5.orig/drivers/video/console/promcon.c
+++ linux-2.6.21-rc5/drivers/video/console/promcon.c
@@ -548,7 +548,8 @@ promcon_scroll(struct vc_data *conp, int
 }
 
 #if !(PROMCON_COLOR)
-static u8 promcon_build_attr(struct vc_data *conp, u8 _color, u8 _intensity, u8 _blink, u8 _underline, u8 _reverse)
+static u8 promcon_build_attr(struct vc_data *conp, u8 _color, u8 _intensity,
+    u8 _blink, u8 _underline, u8 _reverse, u8 _italic)
 {
 	return (_reverse) ? 0xf : 0x7;
 }
Index: linux-2.6.21-rc5/drivers/video/console/sticon.c
===================================================================
--- linux-2.6.21-rc5.orig/drivers/video/console/sticon.c
+++ linux-2.6.21-rc5/drivers/video/console/sticon.c
@@ -314,7 +314,7 @@ static unsigned long sticon_getxy(struct
 }
 
 static u8 sticon_build_attr(struct vc_data *conp, u8 color, u8 intens,
-			    u8 blink, u8 underline, u8 reverse)
+			    u8 blink, u8 underline, u8 reverse, u8 italic)
 {
     u8 attr = ((color & 0x70) >> 1) | ((color & 7));
 
Index: linux-2.6.21-rc5/drivers/video/console/vgacon.c
===================================================================
--- linux-2.6.21-rc5.orig/drivers/video/console/vgacon.c
+++ linux-2.6.21-rc5/drivers/video/console/vgacon.c
@@ -87,7 +87,7 @@ static void vgacon_save_screen(struct vc
 static int vgacon_scroll(struct vc_data *c, int t, int b, int dir,
 			 int lines);
 static u8 vgacon_build_attr(struct vc_data *c, u8 color, u8 intensity,
-			    u8 blink, u8 underline, u8 reverse);
+			    u8 blink, u8 underline, u8 reverse, u8);
 static void vgacon_invert_region(struct vc_data *c, u16 * p, int count);
 static unsigned long vgacon_uni_pagedir[2];
 
@@ -577,12 +577,14 @@ static void vgacon_deinit(struct vc_data
 }
 
 static u8 vgacon_build_attr(struct vc_data *c, u8 color, u8 intensity,
-			    u8 blink, u8 underline, u8 reverse)
+			    u8 blink, u8 underline, u8 reverse, u8 italic)
 {
 	u8 attr = color;
 
 	if (vga_can_do_color) {
-		if (underline)
+		if (italic)
+			attr = (attr & 0xF0) | c->vc_itcolor;
+		else if (underline)
 			attr = (attr & 0xf0) | c->vc_ulcolor;
 		else if (intensity == 0)
 			attr = (attr & 0xf0) | c->vc_halfcolor;
@@ -596,7 +598,9 @@ static u8 vgacon_build_attr(struct vc_da
 	if (intensity == 2)
 		attr ^= 0x08;
 	if (!vga_can_do_color) {
-		if (underline)
+		if (italic)
+			attr = (attr & 0xF8) | 0x02;
+		else if (underline)
 			attr = (attr & 0xf8) | 0x01;
 		else if (intensity == 0)
 			attr = (attr & 0xf0) | 0x08;
Index: linux-2.6.21-rc5/include/linux/console.h
===================================================================
--- linux-2.6.21-rc5.orig/include/linux/console.h
+++ linux-2.6.21-rc5/include/linux/console.h
@@ -51,7 +51,7 @@ struct consw {
 	int	(*con_scrolldelta)(struct vc_data *, int);
 	int	(*con_set_origin)(struct vc_data *);
 	void	(*con_save_screen)(struct vc_data *);
-	u8	(*con_build_attr)(struct vc_data *, u8, u8, u8, u8, u8);
+	u8	(*con_build_attr)(struct vc_data *, u8, u8, u8, u8, u8, u8);
 	void	(*con_invert_region)(struct vc_data *, u16 *, int);
 	u16    *(*con_screen_pos)(struct vc_data *, int);
 	unsigned long (*con_getxy)(struct vc_data *, unsigned long, int *, int *);
Index: linux-2.6.21-rc5/include/linux/console_struct.h
===================================================================
--- linux-2.6.21-rc5.orig/include/linux/console_struct.h
+++ linux-2.6.21-rc5/include/linux/console_struct.h
@@ -37,6 +37,7 @@ struct vc_data {
 	unsigned char	vc_color;		/* Foreground & background */
 	unsigned char	vc_s_color;		/* Saved foreground & background */
 	unsigned char	vc_ulcolor;		/* Color for underline mode */
+	unsigned char   vc_itcolor;
 	unsigned char	vc_halfcolor;		/* Color for half intensity mode */
 	/* cursor */
 	unsigned int	vc_cursor_type;
@@ -71,10 +72,12 @@ struct vc_data {
 	unsigned int	vc_deccolm	: 1;	/* 80/132 Column Mode */
 	/* attribute flags */
 	unsigned int	vc_intensity	: 2;	/* 0=half-bright, 1=normal, 2=bold */
+	unsigned int    vc_italic:1;
 	unsigned int	vc_underline	: 1;
 	unsigned int	vc_blink	: 1;
 	unsigned int	vc_reverse	: 1;
 	unsigned int	vc_s_intensity	: 2;	/* saved rendition */
+	unsigned int    vc_s_italic:1;
 	unsigned int	vc_s_underline	: 1;
 	unsigned int	vc_s_blink	: 1;
 	unsigned int	vc_s_reverse	: 1;
#<EOF>

Re: [PATCH 3/16] vt-underline-color.diff

[Andrew Morton]

On Sun, 1 Apr 2007 20:14:14 +0200 (MEST) Jan Engelhardt <jengelh@linux01.gwdg.de> wrote:

>  static u8 vgacon_build_attr(struct vc_data *c, u8 color, u8 intensity,
> -			    u8 blink, u8 underline, u8 reverse);
> +			    u8 blink, u8 underline, u8 reverse, u8);

ug.  It's better to include the names of the parameters, IMO.  But mixing
the two styles like this is nuts.

Happily, the declaration is unneeded anyway, so I'll just delete it.

Re: [PATCH 3/16] vt-underline-color.diff

[Antonino A. Daplas]

On Sun, 2007-04-01 at 20:14 +0200, Jan Engelhardt wrote:
> Add color support to the "underline" and "italic" attributes as in
> OpenBSD/NetBSD-style (vt220) and xterm.

The italic attribute is indeed part of ANSI standard, though not widely
implemented. Fbcon (monochrome mode) already supports inverse,
underline, and bold. I'll try to implement italics (if I can) too if
this gets in.

> Signed-off-by: Jan Engelhardt <jengelh@gmx.de>
Acked-by: Antonino Daplas <adaplas@gmail.com>

Tony

[PATCH 4/16] vt-printk-color.diff

[Jan Engelhardt]

Colored kernel message output

Gives kernel messages on the console an OpenBSD-esque color to distinguish
them from regular tty (as long as that's not colored too) text.

Inspired by cko (iirc http://freshmeat.net/p/cko/), but better
implementation done by myself (handles newlines correctly).

Signed-off-by: Jan Engelhardt <jengelh@gmx.de>

 Kconfig |   28 ++++++++++++++++++++++++++++
 vt.c    |   22 ++++++++++++++++++++++
 2 files changed, 50 insertions(+)

Index: linux-2.6.21-rc5/drivers/char/Kconfig
===================================================================
--- linux-2.6.21-rc5.orig/drivers/char/Kconfig
+++ linux-2.6.21-rc5/drivers/char/Kconfig
@@ -57,6 +57,34 @@ config VT_CONSOLE
 
 	  If unsure, say Y.
 
+config VT_CKO
+	hex "Colored kernel message output"
+	range 0x00 0xFF
+	depends on VT_CONSOLE
+	default 0x17
+	---help---
+	This option will give you ability to change the color of
+	kernel messages.
+
+	The value you need to enter here is the ASCII color value
+	composed (OR'ed) by one foreground color, one background
+	color and any number of attributes as follows:
+
+	Foreground:     Background:    Attributes:
+	0x01 blue       0x10 blue      0x08 highlight foreground
+	0x02 green      0x20 green     0x80 blinking foreground
+	0x03 cyan       0x30 cyan
+	0x04 red        0x40 red
+	0x05 magenta    0x50 red
+	0x06 brown      0x60 brown
+	0x07 gray       0x70 gray
+
+	Thus, 0x17 will yield gray-on-blue like you can see in
+	OpenBSD, 0x02 the green-on-black like on NetBSD.
+	Using "highlight foreground" does not work when you use
+	VGA Console and fonts with 512 glyphs.
+	VGA Framebuffer is not affected.
+
 config HW_CONSOLE
 	bool
 	depends on VT && !S390 && !UML
Index: linux-2.6.21-rc5/drivers/char/vt.c
===================================================================
--- linux-2.6.21-rc5.orig/drivers/char/vt.c
+++ linux-2.6.21-rc5/drivers/char/vt.c
@@ -73,6 +73,7 @@
  */
 
 #include <linux/module.h>
+#include <linux/moduleparam.h>
 #include <linux/types.h>
 #include <linux/sched.h>
 #include <linux/tty.h>
@@ -2234,6 +2235,17 @@ struct tty_driver *console_driver;
 
 #ifdef CONFIG_VT_CONSOLE
 
+static unsigned int cko_color = CONFIG_VT_CKO;
+
+static __init int set_cko(char *str)
+{
+	cko_color = simple_strtoul(str, NULL, 0);
+	return 1;
+}
+
+__setup("cko=", set_cko);
+module_param(cko_color, uint, S_IRUGO | S_IWUSR);
+
 /*
  *	Console on virtual terminal
  *
@@ -2274,12 +2286,16 @@ static void vt_console_print(struct cons
 		hide_cursor(vc);
 
 	start = (ushort *)vc->vc_pos;
+	vc->vc_color = cko_color;
+	update_attr(vc);
 
 	/* Contrived structure to try to emulate original need_wrap behaviour
 	 * Problems caused when we have need_wrap set on '\n' character */
 	while (count--) {
 		c = *b++;
 		if (c == 10 || c == 13 || c == 8 || vc->vc_need_wrap) {
+			vc->vc_color = vc->vc_def_color;
+			update_attr(vc);
 			if (cnt > 0) {
 				if (CON_IS_VISIBLE(vc))
 					vc->vc_sw->con_putcs(vc, start, cnt, vc->vc_y, vc->vc_x);
@@ -2292,6 +2308,8 @@ static void vt_console_print(struct cons
 				bs(vc);
 				start = (ushort *)vc->vc_pos;
 				myx = vc->vc_x;
+				vc->vc_color = cko_color;
+				update_attr(vc);
 				continue;
 			}
 			if (c != 13)
@@ -2299,6 +2317,8 @@ static void vt_console_print(struct cons
 			cr(vc);
 			start = (ushort *)vc->vc_pos;
 			myx = vc->vc_x;
+			vc->vc_color = cko_color;
+			update_attr(vc);
 			if (c == 10 || c == 13)
 				continue;
 		}
@@ -2320,6 +2340,8 @@ static void vt_console_print(struct cons
 			vc->vc_need_wrap = 1;
 		}
 	}
+	vc->vc_color = vc->vc_def_color;
+	update_attr(vc);
 	set_cursor(vc);
 
 quit:
#<EOF>

[PATCH 5/16] fix-kthread-niceness.diff

[Jan Engelhardt]

Fix kevent's childs priority greedy-ness. Such tasks were always scheduled
at nice level -5 and, at that time, udev stole us the CPU time with -5.

Already posted at http://lkml.org/lkml/2005/1/10/85

Signed-off-by: Jan Engelhardt <jengelh@gmx.de>
Cc: Chris Wright <chrisw@sous-sol.org>

 kmod.c |    1 +
 1 file changed, 1 insertion(+)

Index: linux-2.6.21-rc5/kernel/kmod.c
===================================================================
--- linux-2.6.21-rc5.orig/kernel/kmod.c
+++ linux-2.6.21-rc5/kernel/kmod.c
@@ -165,6 +165,7 @@ static int ____call_usermodehelper(void 
 
 	/* We can run anywhere, unlike our parent keventd(). */
 	set_cpus_allowed(current, CPU_MASK_ALL);
+	set_user_nice(current, 0);
 
 	retval = -EPERM;
 	if (current->fs->root)
#<EOF>

Re: [PATCH 5/16] fix-kthread-niceness.diff

[Andrew Morton]

On Sun, 1 Apr 2007 20:14:40 +0200 (MEST) Jan Engelhardt <jengelh@linux01.gwdg.de> wrote:

> Fix kevent's childs priority greedy-ness. Such tasks were always scheduled
> at nice level -5 and, at that time, udev stole us the CPU time with -5.
> 
> Already posted at http://lkml.org/lkml/2005/1/10/85
> 
> Signed-off-by: Jan Engelhardt <jengelh@gmx.de>
> Cc: Chris Wright <chrisw@sous-sol.org>
> 
>  kmod.c |    1 +
>  1 file changed, 1 insertion(+)
> 
> Index: linux-2.6.21-rc5/kernel/kmod.c
> ===================================================================
> --- linux-2.6.21-rc5.orig/kernel/kmod.c
> +++ linux-2.6.21-rc5/kernel/kmod.c
> @@ -165,6 +165,7 @@ static int ____call_usermodehelper(void 
>  
>  	/* We can run anywhere, unlike our parent keventd(). */
>  	set_cpus_allowed(current, CPU_MASK_ALL);
> +	set_user_nice(current, 0);
>  
>  	retval = -EPERM;
>  	if (current->fs->root)
> #<EOF>

-ETOOOBSCURE.

--- a/kernel/kmod.c~fix-kthread-nicenessdiff-fix
+++ a/kernel/kmod.c
@@ -163,6 +163,11 @@ static int ____call_usermodehelper(void 
 
 	/* We can run anywhere, unlike our parent keventd(). */
 	set_cpus_allowed(current, CPU_MASK_ALL);
+
+	/*
+	 * Our parent is keventd, which runs with elevated scheduling priority.
+	 * Avoid propagating that into the userspace child.
+	 */
 	set_user_nice(current, 0);
 
 	retval = -EPERM;
_

[PATCH 06/16] isofs-add-write-bit.diff

[Jan Engelhardt]

Add %S_IWUGO bit for isofs. Makes it easier to poke on files after
they have been copied from an ISO-9660 filesystem because they are
ofte copied in `cp -p` or perhaps `rsync -p` mode.
They are still not writable though, and isofs will continue to
correctly return -EROFS for writes.

Signed-off-by: Jan Engelhardt <jengelh@gmx.de>

 inode.c |    4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

Index: linux-2.6.21-rc5/fs/isofs/inode.c
===================================================================
--- linux-2.6.21-rc5.orig/fs/isofs/inode.c
+++ linux-2.6.21-rc5/fs/isofs/inode.c
@@ -362,7 +362,7 @@ static int parse_options(char *options, 
 	popt->check = 'u';		/* unset */
 	popt->nocompress = 0;
 	popt->blocksize = 1024;
-	popt->mode = S_IRUGO | S_IXUGO; /* r-x for all.  The disc could
+	popt->mode = S_IRUGO | S_IWUSR | S_IXUGO; /* rwx for all.  The disc could
 					   be shared with DOS machines so
 					   virtually anything could be
 					   a valid executable. */
@@ -1221,7 +1221,7 @@ static void isofs_read_inode(struct inod
 	ei->i_file_format = isofs_file_normal;
 
 	if (de->flags[-high_sierra] & 2) {
-		inode->i_mode = S_IRUGO | S_IXUGO | S_IFDIR;
+		inode->i_mode = S_IRUGO | S_IWUSR | S_IXUGO | S_IFDIR;
 		inode->i_nlink = 1; /* Set to 1.  We know there are 2, but
 				       the find utility tries to optimize
 				       if it is 2, and it screws up.  It is
#<EOF>

[PATCH 07/16] kconfig-dynamic-frequency.diff

[Jan Engelhardt]

Make HZ fully configurable through menuconfig.

Already posted at http://lkml.org/lkml/2006/6/18/111

Signed-off-by: Jan Engelhardt
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk> (http://lkml.org/lkml/2006/11/8/307)
Cc: Sam Ravnborg <sam@ravnborg.org>

 Kconfig.hz |   70 +++++++++++++++++++++++--------------------------------------
 1 file changed, 27 insertions(+), 43 deletions(-)

Index: linux-2.6.21-rc5/kernel/Kconfig.hz
===================================================================
--- linux-2.6.21-rc5.orig/kernel/Kconfig.hz
+++ linux-2.6.21-rc5/kernel/Kconfig.hz
@@ -2,50 +2,34 @@
 # Timer Interrupt Frequency Configuration
 #
 
-choice
-	prompt "Timer frequency"
-	default HZ_250
-	help
-	 Allows the configuration of the timer frequency. It is customary
-	 to have the timer interrupt run at 1000 Hz but 100 Hz may be more
-	 beneficial for servers and NUMA systems that do not need to have
-	 a fast response for user interaction and that may experience bus
-	 contention and cacheline bounces as a result of timer interrupts.
-	 Note that the timer interrupt occurs on each processor in an SMP
-	 environment leading to NR_CPUS * HZ number of timer interrupts
-	 per second.
-
-
-	config HZ_100
-		bool "100 HZ"
-	help
-	  100 Hz is a typical choice for servers, SMP and NUMA systems
-	  with lots of processors that may show reduced performance if
-	  too many timer interrupts are occurring.
-
-	config HZ_250
-		bool "250 HZ"
-	help
-	 250 Hz is a good compromise choice allowing server performance
-	 while also showing good interactive responsiveness even
-	 on SMP and NUMA systems. If you are going to be using NTSC video
-	 or multimedia, selected 300Hz instead.
-
-	config HZ_300
-		bool "300 HZ"
-	help
-	 300 Hz is a good compromise choice allowing server performance
-	 while also showing good interactive responsiveness even
-	 on SMP and NUMA systems and exactly dividing by both PAL and
-	 NTSC frame rates for video and multimedia work.
-
-	config HZ_1000
-		bool "1000 HZ"
-	help
-	 1000 Hz is the preferred choice for desktop systems and other
-	 systems requiring fast interactive responses to events.
+config HZ
+    int "Timer frequency"
+    default 100
+    ---help---
+        Allows the configuration of the timer frequency. It is customary
+        to have the timer interrupt run at 1000 Hz but 100 Hz may be more
+        beneficial for servers and NUMA systems that do not need to have
+        a fast response for user interaction and that may experience bus
+        contention and cacheline bounces as a result of timer interrupts.
+        Note that the timer interrupt occurs on each processor in an SMP
+        environment leading to NR_CPUS * HZ number of timer interrupts
+        per second.
+
+        100 Hz is a typical choice for servers, SMP and NUMA systems
+        with lots of processors that may show reduced performance if
+        too many timer interrupts are occurring.
+
+        250 Hz is a good compromise choice allowing server performance
+        while also showing good interactive responsiveness even on SMP
+        and NUMA systems.
+
+        300 Hz is a good compromise choice allowing server performance
+        while also showing good interactive responsiveness even
+        on SMP and NUMA systems and exactly dividing by both PAL and
+        NTSC frame rates for video and multimedia work.
 
-endchoice
+        1000 Hz is the preferred choice for desktop systems and other
+        systems requiring fast interactive responses to events.
 
 config HZ
 	int
#<EOF>

Re: [PATCH 07/16] kconfig-dynamic-frequency.diff

[Kyle Moffett]

On Apr 01, 2007, at 14:15:20, Jan Engelhardt wrote:
>
> Make HZ fully configurable through menuconfig.
>
> Already posted at http://lkml.org/lkml/2006/6/18/111

Instead of that, what about something like this instead:  NOTE: I  
know my mailer mangles diffs and this one was hand-edited based on  
Jan's diff from within my mailer so it probably doesn't apply, but  
it's enough to point out the idea:

diff --fast -Ndpru linux-2.6.17-rc6~/kernel/Kconfig.hz linux-2.6.17- 
rc6+/kernel/Kconfig.hz
--- linux-2.6.17-rc6~/kernel/Kconfig.hz	2006-06-06 02:57:02.000000000  
+0200
+++ linux-2.6.17-rc6+/kernel/Kconfig.hz	2006-06-16 17:15:46.884794000  
+0200
@@ -36,10 +36,17 @@
  	 1000 HZ is the preferred choice for desktop systems and other
  	 systems requiring fast interactive responses to events.

+	config HZ_CUSTOM
+		bool "Custom HZ"
+	help
+	 This allows you to specify your own custom timer frequency, if
+	 the default options are not satisfactory for your use.
+
  endchoice

  config HZ
-	int
+	int if !HZ_CUSTOM
+	int "Customized HZ value" if HZ_CUSTOM
  	default 100 if HZ_100
  	default 250 if HZ_250
  	default 1000 if HZ_1000

Re: [PATCH 07/16] kconfig-dynamic-frequency.diff

[Jan Engelhardt]

On Apr 1 2007 14:39, Kyle Moffett wrote:
>> 
>> Make HZ fully configurable through menuconfig.
>> 
>> Already posted at http://lkml.org/lkml/2006/6/18/111
>
> Instead of that, what about something like this instead:  NOTE: I know my
> mailer mangles diffs and this one was hand-edited based on Jan's diff from
> within my mailer so it probably doesn't apply, but it's enough to point out the
> idea:

> config HZ
> -	int
> +	int if !HZ_CUSTOM
> +	int "Customized HZ value" if HZ_CUSTOM
> default 100 if HZ_100
> default 250 if HZ_250
> default 1000 if HZ_1000

People always come up with new complexities I never thought kconfig
would be able to do. Nice stuff. I'll take up on your idea and
incorporate it right away.

But I would have to ask: Why do we need 100/250/300/1000 if there is
anyway an input field for the user to enter - s/he should read the
help text for Known Good values, should not s/he?


Jan
-- 

Re: [PATCH 07/16] kconfig-dynamic-frequency.diff

[Kyle Moffett]

On Apr 01, 2007, at 14:42:59, Jan Engelhardt wrote:
> But I would have to ask: Why do we need 100/250/300/1000 if there  
> is anyway an input field for the user to enter - s/he should read  
> the help text for Known Good values, should not s/he?

Actually, probably what should be done is make the "HZ_CUSTOM" option  
depend on EXPERIMENTAL/ 
EMBEDDED/"EXPERT"/"I_KNOW_WHAT_THE_HELL_IM_DOING" or some other  
equivalent option, because a wrong choice for HZ has the potential to  
really screw up a system.  For example, 10000+ would make most  
systems spend a significant portion of their time processing timer  
interrupts and context switches, and less than 100 or so would make  
the system seem jerky.  The point is to encourage the average kernel  
builder not to change from one of the default options unless they  
really know what they're doing.

Cheers,
Kyle Moffett

Re: [PATCH 07/16] kconfig-dynamic-frequency.diff

[Jan Engelhardt]

On Apr 1 2007 14:52, Kyle Moffett wrote:
>
> On Apr 01, 2007, at 14:42:59, Jan Engelhardt wrote:
>> But I would have to ask: Why do we need 100/250/300/1000 if there
>> is anyway an input field for the user to enter - s/he should read
>> the help text for Known Good values, should not s/he?
>
> Actually, probably what should be done is make the "HZ_CUSTOM"
> option depend on EXPERIMENTAL / EMBEDDED / "EXPERT" /
> "I_KNOW_WHAT_THE_HELL_IM_DOING" or some other equivalent option,

EXPERIMENTAL is most often enabled by distributors.
And we are still missing our much-needed CONFIG_EXPERT :-(

> because a wrong choice for HZ has the potential to really screw up
> a system.  For example, 10000+ would make most systems spend a
> significant portion of their time processing timer interrupts and

Point taken. Ehm, I already once tried making a 10000-Hz kernel
around 2.6.13-15 or so. It did not compile because some header files had
#error directives for HZ >= 1536. (jiffies.h:33 if you want
to know.) IIRC there also was another place somewhere in networking
code that #errors out at >8192.


Jan
-- 

[PATCH]

[Kyle Moffett]

[-- Attachment #1: Type: text/plain, Size: 329 bytes --]

Add a CONFIG_I_KNOW_WHAT_THE_HELL_I_AM_DOING variable with a really  
strong
warning attached to it (and a shorter CONFIG_EXPERT variable for use in
Kconfig files where it's a PITA to type out the full name)

Signed-off-by: Kyle Moffett <mrmacman_g4@mac.com>

---

Since my mailer tends to mangle patches I've attached this one


[-- Attachment #2: config_expert.patch.txt --]
[-- Type: text/plain, Size: 1609 bytes --]

Index: linux-2.6.20.1/init/Kconfig
===================================================================
--- linux-2.6.20.1.orig/init/Kconfig
+++ linux-2.6.20.1/init/Kconfig
@@ -40,6 +40,35 @@
 	you say Y here, you will be offered the choice of using features or
 	drivers that are currently considered to be in the alpha-test phase.
 
+config I_KNOW_WHAT_THE_HELL_I_AM_DOING
+	bool "Prompt for config options which will break your computer"
+	---help---
+	  Some kernel developers sometimes need to tune kernel parameters
+	  outside of their normal limits.  Without complete knowledge of
+	  exactly what you are doing you are likely to corrupt the laws of
+	  physics and destroy the universe (well, not quite, but you'll
+	  certainly corrupt your data and crash your box).  If you really
+	  know what you're doing and accept that if it breaks you get to
+	  keep the pieces, then go ahead and enable this option.
+	  
+	  NOTICE:  Bug reports where this option is turned on in the kernel
+	  config will be: (not necessarily in this order)
+	    1)  Laughed at hysterically
+	    2)  Made much fun of
+	    3)  Ridiculed until we're red in the face
+	    4)  Ignored
+	  
+	  WARNING:  If you are building a kernel for a distribution, DO NOT
+	  TURN THIS OPTION ON!!!  Instead you should submit a patch which
+	  adds a new safe value to the list of possibilities for the config
+	  option you're interested in.  Otherwise your users might have a
+	  hard time getting kernel support on the LKML.
+
+config EXPERT
+	bool
+	depends on I_KNOW_WHAT_THE_HELL_I_AM_DOING
+	default y
+
 config BROKEN
 	bool
 

Re: [PATCH]

[Jan Engelhardt]

On Apr 1 2007 15:42, Kyle Moffett wrote:

>Date: Sun, 1 Apr 2007 15:42:09 -0400
>From: Kyle Moffett <mrmacman_g4@mac.com>
>To: Jan Engelhardt <jengelh@linux01.gwdg.de>
>Cc: Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
>    Alan Cox <alan@lxorguk.ukuu.org.uk>, Sam Ravnborg <sam@ravnborg.org>,
>    Andrew Morton <akpm@osdl.org>
>Subject: [PATCH]
>
> Add a CONFIG_I_KNOW_WHAT_THE_HELL_I_AM_DOING variable with a really strong
> warning attached to it (and a shorter CONFIG_EXPERT variable for use in
> Kconfig files where it's a PITA to type out the full name)

Do it like Linus Torvalds and Marissa Mayer and get a better mailer :-D

Index: linux-2.6.20.1/init/Kconfig
===================================================================
--- linux-2.6.20.1.orig/init/Kconfig
+++ linux-2.6.20.1/init/Kconfig
@@ -40,6 +40,35 @@
 	you say Y here, you will be offered the choice of using features or
 	drivers that are currently considered to be in the alpha-test phase.
 
+config I_KNOW_WHAT_THE_HELL_I_AM_DOING

Let's just hope kconfig does not bust the stack.
Perhaps we could shorten that: I_KNOW_WTF_I_AM_DOING

+	bool "Prompt for config options which will break your computer"

s/will break (.*)/could break $1 if done wrong/;

+	---help---
+	  Some kernel developers sometimes need to tune kernel parameters
+	  outside of their normal limits.  Without complete knowledge of
+	  exactly what you are doing you are likely to corrupt the laws of
+	  physics and destroy the universe (well, not quite, but you'll
+	  certainly corrupt your data and crash your box).  If you really
+	  know what you're doing and accept that if it breaks you get to
+	  keep the pieces, then go ahead and enable this option.
+	  
+	  NOTICE:  Bug reports where this option is turned on in the kernel
+	  config will be: (not necessarily in this order)
+	    1)  Laughed at hysterically
+	    2)  Made much fun of
+	    3)  Ridiculed until we're red in the face
+	    4)  Ignored
+	  
+	  WARNING:  If you are building a kernel for a distribution, DO NOT
+	  TURN THIS OPTION ON!!!  Instead you should submit a patch which
+	  adds a new safe value to the list of possibilities for the config
+	  option you're interested in.  Otherwise your users might have a
+	  hard time getting kernel support on the LKML.
+
+config EXPERT
+	bool
+	depends on I_KNOW_WHAT_THE_HELL_I_AM_DOING
+	default y
+
 config BROKEN
 	bool

Given that, I hope the number of options dependent on CONFIG_EXPERT will 
outnumber those depending on CONFIG_I_KNOW_WTF_I_AM_DOING.

If you want my answer: not-so-ACK, move the text to CONFIG_EXPERT and 
let's stay sane for the next 366 days until next April 01.



Jan
-- 

Re: [PATCH]

[Kyle Moffett]

On Apr 01, 2007, at 15:47:18, Jan Engelhardt wrote:
> +config I_KNOW_WHAT_THE_HELL_I_AM_DOING
>
> Let's just hope kconfig does not bust the stack.
> Perhaps we could shorten that: I_KNOW_WTF_I_AM_DOING

Ah, good point, but I think it fits.

> +	bool "Prompt for config options which will break your computer"
>
> s/will break (.*)/could break $1 if done wrong/;

Well, we've had problems in the past where people turn on options  
under CONFIG_EXPERIMENTAL or CONFIG_EMBEDDED and it breaks in exactly  
the specified fashion and they complain on LKML.  The point of this  
is to expose kconfig tunables which a "normal" kconfig user would get  
wrong 99% of the time and would break in obscure ways.  For example,  
the "CONFIG_MULTITHREAD_PCI_PROBE" option, if it was re-added for  
more tinkering, could go under this option.

> +config EXPERT
> +	bool
> +	depends on I_KNOW_WHAT_THE_HELL_I_AM_DOING
> +	default y
>
> Given that, I hope the number of options dependent on CONFIG_EXPERT  
> will outnumber those depending on CONFIG_I_KNOW_WTF_I_AM_DOING.

Ah, see, due to the magic of kconfig language this does not show up  
as a separate option (no text for the "bool") identifier.  It's just  
an alias for I_KNOW_WHAT_THE_HELL_I_AM_DOING to make it easier to use  
in other Kconfig files ("depends EXPERT && SOME_OPTION!=n &&  
ANOTHER_OPTION") instead of s/EXPERT/I_KNOW_WHAT_THE_HELL_I_AM_DOING/g.

> If you want my answer: not-so-ACK, move the text to CONFIG_EXPERT  
> and let's stay sane for the next 366 days until next April 01.

:-D

Cheers,
Kyle Moffett

Re: [PATCH] Add a CONFIG_I_KNOW_WHAT_THE_HELL_I_AM_DOING variable

[Kyle Moffett]

On Apr 01, 2007, at 15:42:09, Kyle Moffett wrote:
> Since my mailer tends to mangle patches I've attached this one

Gah, I complain about my mailer mangling patches and then it's me  
that mangles the patch.  Sorry for forgetting a subject line.

Cheers,
Kyle Moffett

Re: [PATCH]

[Andi Kleen]

Kyle Moffett <mrmacman_g4@mac.com> writes:

> Add a CONFIG_I_KNOW_WHAT_THE_HELL_I_AM_DOING variable with a really
> strong
> warning attached to it (and a shorter CONFIG_EXPERT variable for use in
> Kconfig files where it's a PITA to type out the full name)

When you know what you're doing you can just patch then Kconfig files.
IMHO that's better than this.

-Andi

Re: [PATCH 07/16] kconfig-dynamic-frequency.diff

[Robert P. J. Day]

On Sun, 1 Apr 2007, Kyle Moffett wrote:

> On Apr 01, 2007, at 14:42:59, Jan Engelhardt wrote:

> ... The point is to encourage the average kernel builder not to
> change from one of the default options unless they really know what
> they're doing.

conversely, if someone truly wants to set a kernel config option to
some outrageous value, they should have that freedom.  it's called
"playing."

rday
-- 
========================================================================
Robert P. J. Day
Linux Consulting, Training and Annoying Kernel Pedantry
Waterloo, Ontario, CANADA

http://fsdev.net/wiki/index.php?title=Main_Page
========================================================================

[PATCH 08/16] console-printk-level.diff

[Jan Engelhardt]

Allow the printk level to be set using a "conlevel=" parameter.
(Or is there already a different bootoption that does the same thing?)

Signed-off-by: Jan Engelhardt <jengelh@gmx.de>

 printk.c |    9 +++++++++
 1 file changed, 9 insertions(+)

Index: linux-2.6.21-rc5/kernel/printk.c
===================================================================
--- linux-2.6.21-rc5.orig/kernel/printk.c
+++ linux-2.6.21-rc5/kernel/printk.c
@@ -747,6 +747,15 @@ void resume_console(void)
 }
 #endif /* CONFIG_DISABLE_CONSOLE_SUSPEND */
 
+static __init int set_conlevel(char *str)
+{
+	if(*str >= '0' && *str <= '9')
+		*console_printk = *str - '0';
+	return 1;
+}
+
+__setup("conlevel=", set_conlevel);
+
 /**
  * acquire_console_sem - lock the console system for exclusive use.
  *
#<EOF>

Re: [PATCH 08/16] console-printk-level.diff

[Randy Dunlap]

On Sun, 1 Apr 2007 20:15:27 +0200 (MEST) Jan Engelhardt wrote:

> 
> Allow the printk level to be set using a "conlevel=" parameter.
> (Or is there already a different bootoption that does the same thing?)

Yes, there's "loglevel", as well as
"debug", "quiet", and "ignore_loglevel".


> Signed-off-by: Jan Engelhardt <jengelh@gmx.de>
> 
>  printk.c |    9 +++++++++
>  1 file changed, 9 insertions(+)
> 
> Index: linux-2.6.21-rc5/kernel/printk.c
> ===================================================================
> --- linux-2.6.21-rc5.orig/kernel/printk.c
> +++ linux-2.6.21-rc5/kernel/printk.c
> @@ -747,6 +747,15 @@ void resume_console(void)
>  }
>  #endif /* CONFIG_DISABLE_CONSOLE_SUSPEND */
>  
> +static __init int set_conlevel(char *str)
> +{
> +	if(*str >= '0' && *str <= '9')
> +		*console_printk = *str - '0';

s/*console_printk/console_loglevel/


> +	return 1;
> +}
> +
> +__setup("conlevel=", set_conlevel);

s/conlevel/loglevel/

> +
>  /**
>   * acquire_console_sem - lock the console system for exclusive use.
>   *

---
~Randy
*** Remember to use Documentation/SubmitChecklist when testing your code ***

[PATCH 09/16] zlib-decompression-status.diff

[Jan Engelhardt]

Shows some stars during decompression. This is particularly useful on 
old 386s where one asterisk is probably printed every second - making 
the long linux and initramfs uncompress progress more evident.

Not compile tested outside {i386, x86_64} - don't have other machines.

Signed-off-by: Jan Engelhardt <jengelh@gmx.de>

 arch/alpha/boot/misc.c                    |    2 +-
 arch/arm/boot/compressed/misc.c           |    4 ++--
 arch/arm26/boot/compressed/misc.c         |    4 ++--
 arch/cris/arch-v10/boot/compressed/misc.c |    2 +-
 arch/cris/arch-v32/boot/compressed/misc.c |    2 +-
 arch/i386/boot/compressed/misc.c          |    2 +-
 arch/m32r/boot/compressed/misc.c          |    2 +-
 arch/sh/boot/compressed/misc.c            |    2 +-
 arch/sh64/boot/compressed/misc.c          |    2 +-
 arch/x86_64/boot/compressed/misc.c        |    2 +-
 init/do_mounts_rd.c                       |    7 ++++++-
 init/initramfs.c                          |    7 ++++++-
 lib/inflate.c                             |    9 +++++----
 13 files changed, 29 insertions(+), 18 deletions(-)

Index: linux-2.6.21-rc5/arch/alpha/boot/misc.c
===================================================================
--- linux-2.6.21-rc5.orig/arch/alpha/boot/misc.c
+++ linux-2.6.21-rc5/arch/alpha/boot/misc.c
@@ -201,7 +201,7 @@ decompress_kernel(void *output_start,
 
 	makecrc();
 /*	puts("Uncompressing Linux..."); */
-	gunzip();
+	gunzip(NULL);
 /*	puts(" done, booting the kernel.\n"); */
 	return output_ptr;
 }
Index: linux-2.6.21-rc5/arch/arm/boot/compressed/misc.c
===================================================================
--- linux-2.6.21-rc5.orig/arch/arm/boot/compressed/misc.c
+++ linux-2.6.21-rc5/arch/arm/boot/compressed/misc.c
@@ -355,7 +355,7 @@ decompress_kernel(ulg output_start, ulg 
 
 	makecrc();
 	putstr("Uncompressing Linux...");
-	gunzip();
+	gunzip(putstr);
 	putstr(" done, booting the kernel.\n");
 	return output_ptr;
 }
@@ -369,7 +369,7 @@ int main()
 
 	makecrc();
 	putstr("Uncompressing Linux...");
-	gunzip();
+	gunzip(putstr);
 	putstr("done.\n");
 	return 0;
 }
Index: linux-2.6.21-rc5/arch/arm26/boot/compressed/misc.c
===================================================================
--- linux-2.6.21-rc5.orig/arch/arm26/boot/compressed/misc.c
+++ linux-2.6.21-rc5/arch/arm26/boot/compressed/misc.c
@@ -294,7 +294,7 @@ decompress_kernel(ulg output_start, ulg 
 
 	makecrc();
 	puts("Uncompressing Linux...");
-	gunzip();
+	gunzip(puts);
 	puts(" done, booting the kernel.\n");
 	return output_ptr;
 }
@@ -308,7 +308,7 @@ int main()
 
 	makecrc();
 	puts("Uncompressing Linux...");
-	gunzip();
+	gunzip(puts);
 	puts("done.\n");
 	return 0;
 }
Index: linux-2.6.21-rc5/arch/cris/arch-v10/boot/compressed/misc.c
===================================================================
--- linux-2.6.21-rc5.orig/arch/cris/arch-v10/boot/compressed/misc.c
+++ linux-2.6.21-rc5/arch/cris/arch-v10/boot/compressed/misc.c
@@ -267,6 +267,6 @@ decompress_kernel()
 	}
 
 	puts("Uncompressing Linux...\n");
-	gunzip();
+	gunzip(puts);
 	puts("Done. Now booting the kernel.\n");
 }
Index: linux-2.6.21-rc5/arch/cris/arch-v32/boot/compressed/misc.c
===================================================================
--- linux-2.6.21-rc5.orig/arch/cris/arch-v32/boot/compressed/misc.c
+++ linux-2.6.21-rc5/arch/cris/arch-v32/boot/compressed/misc.c
@@ -312,6 +312,6 @@ decompress_kernel()
 	}
 
 	puts("Uncompressing Linux...\n");
-	gunzip();
+	gunzip(puts);
 	puts("Done. Now booting the kernel.\n");
 }
Index: linux-2.6.21-rc5/arch/i386/boot/compressed/misc.c
===================================================================
--- linux-2.6.21-rc5.orig/arch/i386/boot/compressed/misc.c
+++ linux-2.6.21-rc5/arch/i386/boot/compressed/misc.c
@@ -373,7 +373,7 @@ asmlinkage void decompress_kernel(void *
 
 	makecrc();
 	putstr("Uncompressing Linux... ");
-	gunzip();
+	gunzip(putstr);
 	putstr("Ok, booting the kernel.\n");
 	return;
 }
Index: linux-2.6.21-rc5/arch/m32r/boot/compressed/misc.c
===================================================================
--- linux-2.6.21-rc5.orig/arch/m32r/boot/compressed/misc.c
+++ linux-2.6.21-rc5/arch/m32r/boot/compressed/misc.c
@@ -204,6 +204,6 @@ decompress_kernel(int mmu_on, unsigned c
 
 	makecrc();
 	puts("Uncompressing Linux... ");
-	gunzip();
+	gunzip(puts);
 	puts("Ok, booting the kernel.\n");
 }
Index: linux-2.6.21-rc5/arch/sh/boot/compressed/misc.c
===================================================================
--- linux-2.6.21-rc5.orig/arch/sh/boot/compressed/misc.c
+++ linux-2.6.21-rc5/arch/sh/boot/compressed/misc.c
@@ -236,6 +236,6 @@ void decompress_kernel(void)
 
 	makecrc();
 	puts("Uncompressing Linux... ");
-	gunzip();
+	gunzip(puts);
 	puts("Ok, booting the kernel.\n");
 }
Index: linux-2.6.21-rc5/arch/sh64/boot/compressed/misc.c
===================================================================
--- linux-2.6.21-rc5.orig/arch/sh64/boot/compressed/misc.c
+++ linux-2.6.21-rc5/arch/sh64/boot/compressed/misc.c
@@ -218,7 +218,7 @@ void decompress_kernel(void)
 	makecrc();
 	puts("Uncompressing Linux... ");
 	cache_control(CACHE_ENABLE);
-	gunzip();
+	gunzip(puts);
 	puts("\n");
 
 #if 0
Index: linux-2.6.21-rc5/arch/x86_64/boot/compressed/misc.c
===================================================================
--- linux-2.6.21-rc5.orig/arch/x86_64/boot/compressed/misc.c
+++ linux-2.6.21-rc5/arch/x86_64/boot/compressed/misc.c
@@ -351,7 +351,7 @@ int decompress_kernel(struct moveparams 
 
 	makecrc();
 	putstr(".\nDecompressing Linux...");
-	gunzip();
+	gunzip(putstr);
 	putstr("done.\nBooting the kernel.\n");
 	if (high_loaded) close_output_buffer_if_we_run_high(mv);
 	return high_loaded;
Index: linux-2.6.21-rc5/init/do_mounts_rd.c
===================================================================
--- linux-2.6.21-rc5.orig/init/do_mounts_rd.c
+++ linux-2.6.21-rc5/init/do_mounts_rd.c
@@ -393,6 +393,11 @@ static void __init error(char *x)
 	unzip_error = 1;
 }
 
+static inline void putstr(const char *s) {
+    printk("%s", s);
+    return;
+}
+
 static int __init crd_load(int in_fd, int out_fd)
 {
 	int result;
@@ -418,7 +423,7 @@ static int __init crd_load(int in_fd, in
 		return -1;
 	}
 	makecrc();
-	result = gunzip();
+	result = gunzip(putstr);
 	if (unzip_error)
 		result = 1;
 	kfree(inbuf);
Index: linux-2.6.21-rc5/init/initramfs.c
===================================================================
--- linux-2.6.21-rc5.orig/init/initramfs.c
+++ linux-2.6.21-rc5/init/initramfs.c
@@ -441,6 +441,11 @@ static void __init flush_window(void)
 	outcnt = 0;
 }
 
+static inline void putstr(const char *s) {
+    printk("%s", s);
+    return;
+}
+
 static char * __init unpack_to_rootfs(char *buf, unsigned len, int check_only)
 {
 	int written;
@@ -477,7 +482,7 @@ static char * __init unpack_to_rootfs(ch
 		bytes_out = 0;
 		crc = (ulg)0xffffffffL; /* shift register contents */
 		makecrc();
-		gunzip();
+		gunzip(putstr);
 		if (state != Reset)
 			error("junk in gzipped archive");
 		this_header = saved_offset + inptr;
Index: linux-2.6.21-rc5/lib/inflate.c
===================================================================
--- linux-2.6.21-rc5.orig/lib/inflate.c
+++ linux-2.6.21-rc5/lib/inflate.c
@@ -151,7 +151,7 @@ STATIC int INIT inflate_stored OF((void)
 STATIC int INIT inflate_fixed OF((void));
 STATIC int INIT inflate_dynamic OF((void));
 STATIC int INIT inflate_block OF((int *));
-STATIC int INIT inflate OF((void));
+STATIC int INIT inflate OF((void (*)(const char *)));
 
 
 /* The inflate algorithm uses a sliding 32 K byte window on the uncompressed
@@ -983,7 +983,7 @@ STATIC int INIT inflate_block(
 
 
 
-STATIC int INIT inflate(void)
+STATIC int INIT inflate(void (*putstr)(const char *))
 /* decompress an inflated entry */
 {
   int e;                /* last block flag */
@@ -1000,6 +1000,7 @@ STATIC int INIT inflate(void)
   /* decompress until the last block */
   h = 0;
   do {
+    if(putstr != NULL) putstr("*");
     hufts = 0;
     gzip_mark(&ptr);
     if ((r = inflate_block(&e)) != 0) {
@@ -1093,7 +1094,7 @@ makecrc(void)
 /*
  * Do the uncompression!
  */
-static int INIT gunzip(void)
+static int INIT gunzip(void (*putstr)(const char *))
 {
     uch flags;
     unsigned char magic[2]; /* magic header */
@@ -1157,7 +1158,7 @@ static int INIT gunzip(void)
     }
 
     /* Decompress */
-    if ((res = inflate())) {
+    if ((res = inflate(putstr))) {
 	    switch (res) {
 	    case 0:
 		    break;
#<EOF>

Re: [PATCH 09/16] zlib-decompression-status.diff

[Jörn Engel]

On Sun, 1 April 2007 20:15:42 +0200, Jan Engelhardt wrote:
>  
> +static inline void putstr(const char *s) {
> +    printk("%s", s);
> +    return;
> +}
> +
>  static int __init crd_load(int in_fd, int out_fd)
>  {
>  	int result;
> @@ -418,7 +423,7 @@ static int __init crd_load(int in_fd, in
>  		return -1;
>  	}
>  	makecrc();
> -	result = gunzip();
> +	result = gunzip(putstr);

You are sure this wasn't meant as an April fools joke?  Passing the
address of an inline function certainly has a humorous aspect. ;)

Also, you can remove the "return;" in the void function and possibly
change this bit to match Documentation/CodingStyle.

> +    if(putstr != NULL) putstr("*");

The patch alternately uses puts() and putstr(), which looks rather odd.
Not sure whether that makes sense or not.

Jörn

-- 
My second remark is that our intellectual powers are rather geared to
master static relations and that our powers to visualize processes
evolving in time are relatively poorly developed.
-- Edsger W. Dijkstra

Re: [PATCH 09/16] zlib-decompression-status.diff

[Jan Engelhardt]

Hi,


On Apr 2 2007 19:48, Jörn Engel wrote:
>On Sun, 1 April 2007 20:15:42 +0200, Jan Engelhardt wrote:
>>  
>> +static inline void putstr(const char *s) {
>> +    printk("%s", s);
>> +    return;
>> +}
>> +
>>  static int __init crd_load(int in_fd, int out_fd)
>>  {
>>  	int result;
>> @@ -418,7 +423,7 @@ static int __init crd_load(int in_fd, in
>>  		return -1;
>>  	}
>>  	makecrc();
>> -	result = gunzip();
>> +	result = gunzip(putstr);
>
>You are sure this wasn't meant as an April fools joke?

It actually has use (for my 20 MHz 386 *hug*)

>Passing the address of an inline function certainly has a humorous aspect. ;)

You bet. Well, "inline" is, as far as I hear the voices of the C standard, just
a "hint" to the compiler that this function should be inlined. If it cannot do
that, well, then do not inline it. But you have a point, it gets interesting
when "inline" expands to __attribute__((always_inline)). Well, let's hope that
gcc 'carries the inlining over' into gunzip().

>Also, you can remove the "return;" in the void function and possibly
>change this bit to match Documentation/CodingStyle.

Yup.

>> +    if(putstr != NULL) putstr("*");
>
>The patch alternately uses puts() and putstr(), which looks rather odd.

It is because some arches seem to have a arch-specific "output function"
called puts() try the following:

$ grep -r 'puts(' arch/arm26/
(results)
$ grep -r 'puts(' arch/i386/
(no results)

So yes, I would need some advice what people were thinking here. If it
was not for this arch specific stuff, the putstr() function I added in
lib/ would also be gone and the "*" could be directly printed in
gunzip(), making the extra argument to gunzip superfluous.

The extra putstr() in init/initramfs.c [cf. putstr() in lib/inflate.c]
is also something I did not know how to really do better.

Thoughts are appreciated and stylofixed patch below.
(lib/inflate.c does not even use tabs, what could I possibly do more wrong than
already? :p)


---

Shows some stars during decompression. This is particularly useful on old 386s
where one asterisk is probably printed every second - making the long linux and
initramfs uncompress progress more evident.
Not compile tested outside {i386, x86_64} - don't have other machines.

Signed-off-by: Jan Engelhardt <jengelh@gmx.de>

 arch/alpha/boot/misc.c                    |    2 +-
 arch/arm/boot/compressed/misc.c           |    4 ++--
 arch/arm26/boot/compressed/misc.c         |    4 ++--
 arch/cris/arch-v10/boot/compressed/misc.c |    2 +-
 arch/cris/arch-v32/boot/compressed/misc.c |    2 +-
 arch/i386/boot/compressed/misc.c          |    2 +-
 arch/m32r/boot/compressed/misc.c          |    2 +-
 arch/sh/boot/compressed/misc.c            |    2 +-
 arch/sh64/boot/compressed/misc.c          |    2 +-
 arch/x86_64/boot/compressed/misc.c        |    2 +-
 init/do_mounts_rd.c                       |    7 ++++++-
 init/initramfs.c                          |    7 ++++++-
 lib/inflate.c                             |   10 ++++++----
 13 files changed, 30 insertions(+), 18 deletions(-)

Index: linux-2.6.21-rc5/arch/alpha/boot/misc.c
===================================================================
--- linux-2.6.21-rc5.orig/arch/alpha/boot/misc.c
+++ linux-2.6.21-rc5/arch/alpha/boot/misc.c
@@ -201,7 +201,7 @@ decompress_kernel(void *output_start,
 
 	makecrc();
 /*	puts("Uncompressing Linux..."); */
-	gunzip();
+	gunzip(NULL);
 /*	puts(" done, booting the kernel.\n"); */
 	return output_ptr;
 }
Index: linux-2.6.21-rc5/arch/arm/boot/compressed/misc.c
===================================================================
--- linux-2.6.21-rc5.orig/arch/arm/boot/compressed/misc.c
+++ linux-2.6.21-rc5/arch/arm/boot/compressed/misc.c
@@ -355,7 +355,7 @@ decompress_kernel(ulg output_start, ulg 
 
 	makecrc();
 	putstr("Uncompressing Linux...");
-	gunzip();
+	gunzip(putstr);
 	putstr(" done, booting the kernel.\n");
 	return output_ptr;
 }
@@ -369,7 +369,7 @@ int main()
 
 	makecrc();
 	putstr("Uncompressing Linux...");
-	gunzip();
+	gunzip(putstr);
 	putstr("done.\n");
 	return 0;
 }
Index: linux-2.6.21-rc5/arch/arm26/boot/compressed/misc.c
===================================================================
--- linux-2.6.21-rc5.orig/arch/arm26/boot/compressed/misc.c
+++ linux-2.6.21-rc5/arch/arm26/boot/compressed/misc.c
@@ -294,7 +294,7 @@ decompress_kernel(ulg output_start, ulg 
 
 	makecrc();
 	puts("Uncompressing Linux...");
-	gunzip();
+	gunzip(puts);
 	puts(" done, booting the kernel.\n");
 	return output_ptr;
 }
@@ -308,7 +308,7 @@ int main()
 
 	makecrc();
 	puts("Uncompressing Linux...");
-	gunzip();
+	gunzip(puts);
 	puts("done.\n");
 	return 0;
 }
Index: linux-2.6.21-rc5/arch/cris/arch-v10/boot/compressed/misc.c
===================================================================
--- linux-2.6.21-rc5.orig/arch/cris/arch-v10/boot/compressed/misc.c
+++ linux-2.6.21-rc5/arch/cris/arch-v10/boot/compressed/misc.c
@@ -267,6 +267,6 @@ decompress_kernel()
 	}
 
 	puts("Uncompressing Linux...\n");
-	gunzip();
+	gunzip(puts);
 	puts("Done. Now booting the kernel.\n");
 }
Index: linux-2.6.21-rc5/arch/cris/arch-v32/boot/compressed/misc.c
===================================================================
--- linux-2.6.21-rc5.orig/arch/cris/arch-v32/boot/compressed/misc.c
+++ linux-2.6.21-rc5/arch/cris/arch-v32/boot/compressed/misc.c
@@ -312,6 +312,6 @@ decompress_kernel()
 	}
 
 	puts("Uncompressing Linux...\n");
-	gunzip();
+	gunzip(puts);
 	puts("Done. Now booting the kernel.\n");
 }
Index: linux-2.6.21-rc5/arch/i386/boot/compressed/misc.c
===================================================================
--- linux-2.6.21-rc5.orig/arch/i386/boot/compressed/misc.c
+++ linux-2.6.21-rc5/arch/i386/boot/compressed/misc.c
@@ -373,7 +373,7 @@ asmlinkage void decompress_kernel(void *
 
 	makecrc();
 	putstr("Uncompressing Linux... ");
-	gunzip();
+	gunzip(putstr);
 	putstr("Ok, booting the kernel.\n");
 	return;
 }
Index: linux-2.6.21-rc5/arch/m32r/boot/compressed/misc.c
===================================================================
--- linux-2.6.21-rc5.orig/arch/m32r/boot/compressed/misc.c
+++ linux-2.6.21-rc5/arch/m32r/boot/compressed/misc.c
@@ -204,6 +204,6 @@ decompress_kernel(int mmu_on, unsigned c
 
 	makecrc();
 	puts("Uncompressing Linux... ");
-	gunzip();
+	gunzip(puts);
 	puts("Ok, booting the kernel.\n");
 }
Index: linux-2.6.21-rc5/arch/sh/boot/compressed/misc.c
===================================================================
--- linux-2.6.21-rc5.orig/arch/sh/boot/compressed/misc.c
+++ linux-2.6.21-rc5/arch/sh/boot/compressed/misc.c
@@ -236,6 +236,6 @@ void decompress_kernel(void)
 
 	makecrc();
 	puts("Uncompressing Linux... ");
-	gunzip();
+	gunzip(puts);
 	puts("Ok, booting the kernel.\n");
 }
Index: linux-2.6.21-rc5/arch/sh64/boot/compressed/misc.c
===================================================================
--- linux-2.6.21-rc5.orig/arch/sh64/boot/compressed/misc.c
+++ linux-2.6.21-rc5/arch/sh64/boot/compressed/misc.c
@@ -218,7 +218,7 @@ void decompress_kernel(void)
 	makecrc();
 	puts("Uncompressing Linux... ");
 	cache_control(CACHE_ENABLE);
-	gunzip();
+	gunzip(puts);
 	puts("\n");
 
 #if 0
Index: linux-2.6.21-rc5/arch/x86_64/boot/compressed/misc.c
===================================================================
--- linux-2.6.21-rc5.orig/arch/x86_64/boot/compressed/misc.c
+++ linux-2.6.21-rc5/arch/x86_64/boot/compressed/misc.c
@@ -351,7 +351,7 @@ int decompress_kernel(struct moveparams 
 
 	makecrc();
 	putstr(".\nDecompressing Linux...");
-	gunzip();
+	gunzip(putstr);
 	putstr("done.\nBooting the kernel.\n");
 	if (high_loaded) close_output_buffer_if_we_run_high(mv);
 	return high_loaded;
Index: linux-2.6.21-rc5/init/do_mounts_rd.c
===================================================================
--- linux-2.6.21-rc5.orig/init/do_mounts_rd.c
+++ linux-2.6.21-rc5/init/do_mounts_rd.c
@@ -393,6 +393,11 @@ static void __init error(char *x)
 	unzip_error = 1;
 }
 
+static void putstr(const char *s)
+{
+	printk("%s", s);
+}
+
 static int __init crd_load(int in_fd, int out_fd)
 {
 	int result;
@@ -418,7 +423,7 @@ static int __init crd_load(int in_fd, in
 		return -1;
 	}
 	makecrc();
-	result = gunzip();
+	result = gunzip(putstr);
 	if (unzip_error)
 		result = 1;
 	kfree(inbuf);
Index: linux-2.6.21-rc5/init/initramfs.c
===================================================================
--- linux-2.6.21-rc5.orig/init/initramfs.c
+++ linux-2.6.21-rc5/init/initramfs.c
@@ -441,6 +441,11 @@ static void __init flush_window(void)
 	outcnt = 0;
 }
 
+static void putstr(const char *s)
+{
+	printk("%s", s);
+}
+
 static char * __init unpack_to_rootfs(char *buf, unsigned len, int check_only)
 {
 	int written;
@@ -477,7 +482,7 @@ static char * __init unpack_to_rootfs(ch
 		bytes_out = 0;
 		crc = (ulg)0xffffffffL; /* shift register contents */
 		makecrc();
-		gunzip();
+		gunzip(putstr);
 		if (state != Reset)
 			error("junk in gzipped archive");
 		this_header = saved_offset + inptr;
Index: linux-2.6.21-rc5/lib/inflate.c
===================================================================
--- linux-2.6.21-rc5.orig/lib/inflate.c
+++ linux-2.6.21-rc5/lib/inflate.c
@@ -151,7 +151,7 @@ STATIC int INIT inflate_stored OF((void)
 STATIC int INIT inflate_fixed OF((void));
 STATIC int INIT inflate_dynamic OF((void));
 STATIC int INIT inflate_block OF((int *));
-STATIC int INIT inflate OF((void));
+STATIC int INIT inflate OF((void (*)(const char *)));
 
 
 /* The inflate algorithm uses a sliding 32 K byte window on the uncompressed
@@ -983,7 +983,7 @@ STATIC int INIT inflate_block(
 
 
 
-STATIC int INIT inflate(void)
+STATIC int INIT inflate(void (*putstr)(const char *))
 /* decompress an inflated entry */
 {
   int e;                /* last block flag */
@@ -1000,6 +1000,8 @@ STATIC int INIT inflate(void)
   /* decompress until the last block */
   h = 0;
   do {
+    if(putstr != NULL)
+      putstr("*");
     hufts = 0;
     gzip_mark(&ptr);
     if ((r = inflate_block(&e)) != 0) {
@@ -1093,7 +1095,7 @@ makecrc(void)
 /*
  * Do the uncompression!
  */
-static int INIT gunzip(void)
+static int INIT gunzip(void (*putstr)(const char *))
 {
     uch flags;
     unsigned char magic[2]; /* magic header */
@@ -1157,7 +1159,7 @@ static int INIT gunzip(void)
     }
 
     /* Decompress */
-    if ((res = inflate())) {
+    if ((res = inflate(putstr))) {
 	    switch (res) {
 	    case 0:
 		    break;
#<EOF>


Jan
-- 

[PATCH 10/16] show-partitions-on-mount-error.diff

[Jan Engelhardt]

Display all possible partitions when the root filesystem is not mounted.
This helps to track spell'o's and missing drivers.

From: David Alan Gilbert, http://lkml.org/lkml/2005/2/26/92

Updated to work with newer kernels.

Signed-off-by: Jan Engelhardt <jengelh@gmx.de>

 block/genhd.c         |   54 ++++++++++++++++++++++++++++++++++++++++++++++++++
 include/linux/genhd.h |    1 
 init/do_mounts.c      |    7 +++++-
 3 files changed, 61 insertions(+), 1 deletion(-)

Index: linux-2.6.21-rc5/block/genhd.c
===================================================================
--- linux-2.6.21-rc5.orig/block/genhd.c
+++ linux-2.6.21-rc5/block/genhd.c
@@ -215,6 +215,60 @@ struct gendisk *get_gendisk(dev_t dev, i
 	return  kobj ? to_disk(kobj) : NULL;
 }
 
+/*
+ * printk a full list of all partitions - intended for
+ * places where the root filesystem can't be mounted and thus
+ * to give the victim some idea of what went wrong
+ */
+void printk_all_partitions(void)
+{
+	int n;
+        struct gendisk* sgp;
+	mutex_lock(&block_subsys_lock);
+
+        /* For each block device... */
+	list_for_each_entry(sgp, &block_subsys.kset.list, kobj.entry) {
+		char buf[BDEVNAME_SIZE];
+	  	/*
+	  	 * Don't show empty devices or things that have been surpressed
+	  	 */
+		if (get_capacity(sgp) &&
+		  !(sgp->flags & GENHD_FL_SUPPRESS_PARTITION_INFO)) {
+			/*
+			 * Note, unlike /proc/partitions I am showing the
+			 * numbers in hex in the same format as the root=
+			 * option
+			 */
+			printk("%02x%02x %10llu %s",
+			       sgp->major, sgp->first_minor,
+			       (unsigned long long)get_capacity(sgp) >> 1,
+			       disk_name(sgp, 0, buf));
+			if ((sgp->driverfs_dev) &&
+			  (sgp->driverfs_dev->driver)) {
+				printk(" driver: %s\n",
+				       sgp->driverfs_dev->driver->name);
+			} else {
+				printk(" (driver?)\n");
+			}
+
+			/* now show the partitions */
+			for (n = 0; n < sgp->minors - 1; ++n) {
+				if (!sgp->part[n])
+					continue;
+				if (sgp->part[n]->nr_sects == 0)
+					continue;
+				printk("  %02x%02x %10llu %s\n",
+				       sgp->major, n + 1 + sgp->first_minor,
+				       (unsigned long long)sgp->part[n]->nr_sects >> 1 ,
+				       disk_name(sgp, n + 1, buf));
+			} /* partition subloop */
+		} /* Non 0 size, not surpressed */
+	} /* Block device loop */
+
+	mutex_unlock(&block_subsys_lock);
+	return;
+}
+
 #ifdef CONFIG_PROC_FS
 /* iterator */
 static void *part_start(struct seq_file *part, loff_t *pos)
Index: linux-2.6.21-rc5/include/linux/genhd.h
===================================================================
--- linux-2.6.21-rc5.orig/include/linux/genhd.h
+++ linux-2.6.21-rc5/include/linux/genhd.h
@@ -413,6 +413,7 @@ char *disk_name (struct gendisk *hd, int
 extern int rescan_partitions(struct gendisk *disk, struct block_device *bdev);
 extern void add_partition(struct gendisk *, int, sector_t, sector_t, int);
 extern void delete_partition(struct gendisk *, int);
+extern void printk_all_partitions(void);
 
 extern struct gendisk *alloc_disk_node(int minors, int node_id);
 extern struct gendisk *alloc_disk(int minors);
Index: linux-2.6.21-rc5/init/do_mounts.c
===================================================================
--- linux-2.6.21-rc5.orig/init/do_mounts.c
+++ linux-2.6.21-rc5/init/do_mounts.c
@@ -7,6 +7,7 @@
 #include <linux/root_dev.h>
 #include <linux/security.h>
 #include <linux/delay.h>
+#include <linux/genhd.h>
 #include <linux/mount.h>
 #include <linux/device.h>
 
@@ -307,17 +308,21 @@ retry:
 	        /*
 		 * Allow the user to distinguish between failed sys_open
 		 * and bad superblock on root device.
+		 * and give them a list of the available devices
 		 */
 #ifdef CONFIG_BLOCK
 		__bdevname(ROOT_DEV, b);
 #endif
 		printk("VFS: Cannot open root device \"%s\" or %s\n",
 				root_device_name, b);
-		printk("Please append a correct \"root=\" boot option\n");
+		printk("Please append a correct \"root=\" boot option; here are the available partitions:\n");
 
+		printk_all_partitions();
 		panic("VFS: Unable to mount root fs on %s", b);
 	}
 
+	printk("List of all partitions:\n");
+	printk_all_partitions();
 	printk("No filesystem could mount root, tried: ");
 	for (p = fs_names; *p; p += strlen(p)+1)
 		printk(" %s", p);
#<EOF>

Re: [PATCH 10/16] show-partitions-on-mount-error.diff

[Andrew Morton]

On Sun, 1 Apr 2007 20:15:53 +0200 (MEST) Jan Engelhardt <jengelh@linux01.gwdg.de> wrote:

> 
> Display all possible partitions when the root filesystem is not mounted.
> This helps to track spell'o's and missing drivers.
> 
> From: David Alan Gilbert, http://lkml.org/lkml/2005/2/26/92
> 
> Updated to work with newer kernels.
> 
> Signed-off-by: Jan Engelhardt <jengelh@gmx.de>
> 
> +void printk_all_partitions(void)

I made this __init.

Re: [PATCH 10/16] show-partitions-on-mount-error.diff

[Andrew Morton]

On Sun, 1 Apr 2007 20:15:53 +0200 (MEST) Jan Engelhardt <jengelh@linux01.gwdg.de> wrote:

> Display all possible partitions when the root filesystem is not mounted.
> This helps to track spell'o's and missing drivers.

It would be nice to include a sample of the resulting output for people
to review.

Re: [PATCH 10/16] show partitions on mount error

[Jan Engelhardt]

Hi,


On Apr 1 2007 22:47, Andrew Morton wrote:
>On Sun, 1 Apr 2007 20:15:53 +0200 (MEST) Jan Engelhardt <jengelh@linux01.gwdg.de> wrote:
>> 
>> Display all possible partitions when the root filesystem is not mounted.
>> This helps to track spell'o's and missing drivers.
>> 
>> +void printk_all_partitions(void)
>
>I made this __init.

Seems reasonable.
I have done some additional cleanups, see patch below.


>It would be nice to include a sample of the resulting output for people
>to review.

Loaded with PXELINUX...

	LABEL l
		KERNEL bzImage
		APPEND root=/dev/foobar

or root=/dev/sdb both produce:

[...]
VFS: Cannot open root device "foobar" or unknown-block(0,0)
Please append a correct "root=" boot option; here are the available partitions:
0800    8388608 sda driver: sd
  0801     192748 sda1
  0802    8193150 sda2
0810    4194304 sdb driver: sd
Kernel panic - not syncing: VFS: Unable to mount root fs on unknown-block(0,0)

With root=/dev/sdb, the list of possible filesystems is also displayed
(a patch that was merged a (long)while back).


---

Integrated __init annotation, and deindented the loop by one level.
Replaces original patch.

Signed-off-by: Jan Engelhardt <jengelh@gmx.de>

Index: linux-2.6.21-rc5/block/genhd.c
===================================================================
--- linux-2.6.21-rc5.orig/block/genhd.c
+++ linux-2.6.21-rc5/block/genhd.c
@@ -215,6 +215,59 @@ struct gendisk *get_gendisk(dev_t dev, i
 	return  kobj ? to_disk(kobj) : NULL;
 }
 
+/*
+ * printk a full list of all partitions - intended for
+ * places where the root filesystem can't be mounted and thus
+ * to give the victim some idea of what went wrong
+ */
+void __init printk_all_partitions(void)
+{
+	int n;
+	struct gendisk *sgp;
+	mutex_lock(&block_subsys_lock);
+
+	/* For each block device... */
+	list_for_each_entry(sgp, &block_subsys.kset.list, kobj.entry) {
+		char buf[BDEVNAME_SIZE];
+		/*
+		 * Don't show empty devices or things that have been surpressed
+		 */
+		if (get_capacity(sgp) == 0 ||
+		  (sgp->flags & GENHD_FL_SUPPRESS_PARTITION_INFO))
+			continue;
+
+		/*
+		 * Note, unlike /proc/partitions, I am showing the numbers in
+		 * hex - the same format as the root= option takes.
+		 */
+		printk("%02x%02x %10llu %s",
+		       sgp->major, sgp->first_minor,
+		       (unsigned long long)get_capacity(sgp) >> 1,
+		       disk_name(sgp, 0, buf));
+		if (sgp->driverfs_dev != NULL &&
+		  sgp->driverfs_dev->driver != NULL)
+			printk(" driver: %s\n",
+			       sgp->driverfs_dev->driver->name);
+		else
+			printk(" (driver?)\n");
+
+		/* now show the partitions */
+		for (n = 0; n < sgp->minors - 1; ++n) {
+			if (sgp->part[n] == NULL)
+				continue;
+			if (sgp->part[n]->nr_sects == 0)
+				continue;
+			printk("  %02x%02x %10llu %s\n",
+			       sgp->major, n + 1 + sgp->first_minor,
+			       (unsigned long long)sgp->part[n]->nr_sects >> 1,
+			       disk_name(sgp, n + 1, buf));
+		} /* partition subloop */
+	} /* Block device loop */
+
+	mutex_unlock(&block_subsys_lock);
+	return;
+}
+
 #ifdef CONFIG_PROC_FS
 /* iterator */
 static void *part_start(struct seq_file *part, loff_t *pos)
Index: linux-2.6.21-rc5/include/linux/genhd.h
===================================================================
--- linux-2.6.21-rc5.orig/include/linux/genhd.h
+++ linux-2.6.21-rc5/include/linux/genhd.h
@@ -413,6 +413,7 @@ char *disk_name (struct gendisk *hd, int
 extern int rescan_partitions(struct gendisk *disk, struct block_device *bdev);
 extern void add_partition(struct gendisk *, int, sector_t, sector_t, int);
 extern void delete_partition(struct gendisk *, int);
+extern void printk_all_partitions(void);
 
 extern struct gendisk *alloc_disk_node(int minors, int node_id);
 extern struct gendisk *alloc_disk(int minors);
Index: linux-2.6.21-rc5/init/do_mounts.c
===================================================================
--- linux-2.6.21-rc5.orig/init/do_mounts.c
+++ linux-2.6.21-rc5/init/do_mounts.c
@@ -7,6 +7,7 @@
 #include <linux/root_dev.h>
 #include <linux/security.h>
 #include <linux/delay.h>
+#include <linux/genhd.h>
 #include <linux/mount.h>
 #include <linux/device.h>
 
@@ -307,17 +308,21 @@ retry:
 	        /*
 		 * Allow the user to distinguish between failed sys_open
 		 * and bad superblock on root device.
+		 * and give them a list of the available devices
 		 */
 #ifdef CONFIG_BLOCK
 		__bdevname(ROOT_DEV, b);
 #endif
 		printk("VFS: Cannot open root device \"%s\" or %s\n",
 				root_device_name, b);
-		printk("Please append a correct \"root=\" boot option\n");
+		printk("Please append a correct \"root=\" boot option; here are the available partitions:\n");
 
+		printk_all_partitions();
 		panic("VFS: Unable to mount root fs on %s", b);
 	}
 
+	printk("List of all partitions:\n");
+	printk_all_partitions();
 	printk("No filesystem could mount root, tried: ");
 	for (p = fs_names; *p; p += strlen(p)+1)
 		printk(" %s", p);
#<EOF>

Re: [PATCH 10/16] show-partitions-on-mount-error.diff

[Pavel Machek]

Hi!

>  block/genhd.c         |   54 ++++++++++++++++++++++++++++++++++++++++++++++++++
>  include/linux/genhd.h |    1 
>  init/do_mounts.c      |    7 +++++-
>  3 files changed, 61 insertions(+), 1 deletion(-)
> 
> Index: linux-2.6.21-rc5/block/genhd.c
> ===================================================================
> --- linux-2.6.21-rc5.orig/block/genhd.c
> +++ linux-2.6.21-rc5/block/genhd.c
> @@ -215,6 +215,60 @@ struct gendisk *get_gendisk(dev_t dev, i
>  	return  kobj ? to_disk(kobj) : NULL;
>  }
>  
> +/*
> + * printk a full list of all partitions - intended for
> + * places where the root filesystem can't be mounted and thus
> + * to give the victim some idea of what went wrong
> + */
> +void printk_all_partitions(void)
> +{
> +	int n;
> +        struct gendisk* sgp;
> +	mutex_lock(&block_subsys_lock);

tabs-vs-spaces problem.


-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

Re: [PATCH 10/16] show-partitions-on-mount-error.diff

[Jan Engelhardt]

On Apr 2 2007 18:59, Pavel Machek wrote in http://lkml.org/lkml/2007/4/3/85:
>
>> + * to give the victim some idea of what went wrong
>> + */
>> +void printk_all_partitions(void)
>> +{
>> +	int n;
>> +        struct gendisk* sgp;
>> +	mutex_lock(&block_subsys_lock);
>
>tabs-vs-spaces problem.

Already fixed and sent out in http://lkml.org/lkml/2007/4/2/75


Jan
-- 

[PATCH 11/16] samba-eintr-fix.diff

[Jan Engelhardt]

From: Dave Jones
Already posted at http://lkml.org/lkml/2005/7/7/255

Signed-off-by: Jan Engelhardt <jengelh@gmx.de>
Cc: Dave Jones <Dave Jones <davej@redhat.com>

 request.c |    1 +
 1 file changed, 1 insertion(+)

Index: linux-2.6.21-rc5/fs/smbfs/request.c
===================================================================
--- linux-2.6.21-rc5.orig/fs/smbfs/request.c
+++ linux-2.6.21-rc5/fs/smbfs/request.c
@@ -347,6 +347,7 @@ int smb_add_request(struct smb_request *
 			smb_rput(req);
 		}
 		smb_unlock_server(server);
+		return -EINTR;
 	}
 
 	if (!timeleft) {
#<EOF>

Re: [PATCH 11/16] samba-eintr-fix.diff

[Dave Jones]

On Sun, Apr 01, 2007 at 08:16:22PM +0200, Jan Engelhardt wrote:
 > 
 > From: Dave Jones
 > Already posted at http://lkml.org/lkml/2005/7/7/255

I had to stop and think if this was an April 1st :)
Back in 2005 it seemed I had a strong enough stomach
to wade through smbfs for some unknown reason.
These days like many others, I wish someone would
just put the thing out of its misery.

In Fedora we switched it off (in about 2005 iirc)
in favor of using cifs, which for the most part
is a drop-in replacement.  There still remain 1-2
corner cases where it doesn't work iirc, but
the situation is a lot better than it used to be.

As to whether the patch is worthwhile/correct..
I really don't know.  I posted it back then in the hopes
that someone smbfs-savvy would pop up and review it.
As you can see from the archive, no-one did.

	Dave

-- 
http://www.codemonkey.org.uk

Re: [PATCH 11/16] samba-eintr-fix.diff

[Jan Engelhardt]

On Apr 1 2007 15:09, Dave Jones wrote:
>On Sun, Apr 01, 2007 at 08:16:22PM +0200, Jan Engelhardt wrote:

> > From: Dave Jones
> > Already posted at http://lkml.org/lkml/2005/7/7/255
>
>I had to stop and think if this was an April 1st :)

Read the mail entitled "[PATCH 0/16] Assorted patches".

>As to whether the patch is worthwhile/correct..
>I really don't know.  I posted it back then in the hopes
>that someone smbfs-savvy would pop up and review it.
>As you can see from the archive, no-one did.

It seems to have been resent once, 7 days after you:
	http://lkml.org/lkml/2005/7/15/231

At that time, I had experienced strange copy failures ("No such file or 
directory") while the directory being copied was not modified. With the 
patch, I at least got EINTR instead of ENOENT, which I think made cp 
think once more to retry.


Jan
-- 

Re: [PATCH 11/16] samba-eintr-fix.diff

[Dave Jones]

On Sun, Apr 01, 2007 at 09:28:03PM +0200, Jan Engelhardt wrote:
 > 
 > On Apr 1 2007 15:09, Dave Jones wrote:
 > >On Sun, Apr 01, 2007 at 08:16:22PM +0200, Jan Engelhardt wrote:
 > 
 > > > From: Dave Jones
 > > > Already posted at http://lkml.org/lkml/2005/7/7/255
 > >
 > >I had to stop and think if this was an April 1st :)
 > 
 > Read the mail entitled "[PATCH 0/16] Assorted patches".

That's what happens when you read inbox before mailing list folders :)

 > >As to whether the patch is worthwhile/correct..
 > >I really don't know.  I posted it back then in the hopes
 > >that someone smbfs-savvy would pop up and review it.
 > >As you can see from the archive, no-one did.
 > 
 > It seems to have been resent once, 7 days after you:
 > 	http://lkml.org/lkml/2005/7/15/231

Hmm, with no changelog. I wonder if JA was hitting that bug
and independantly came up with that patch, or was just
resubmitting on my behalf.

 > At that time, I had experienced strange copy failures ("No such file or 
 > directory") while the directory being copied was not modified. With the 
 > patch, I at least got EINTR instead of ENOENT, which I think made cp 
 > think once more to retry.

I suppose we could throw it in -mm and see if anyone screams. Andrew?

	Dave

-- 
http://www.codemonkey.org.uk

Re: [PATCH 11/16] samba-eintr-fix.diff

[Andrew Morton]

On Sun, 1 Apr 2007 20:16:22 +0200 (MEST) Jan Engelhardt <jengelh@linux01.gwdg.de> wrote:

> 
> From: Dave Jones
> Already posted at http://lkml.org/lkml/2005/7/7/255
> 
> Signed-off-by: Jan Engelhardt <jengelh@gmx.de>
> Cc: Dave Jones <Dave Jones <davej@redhat.com>
> 
>  request.c |    1 +
>  1 file changed, 1 insertion(+)
> 
> Index: linux-2.6.21-rc5/fs/smbfs/request.c
> ===================================================================
> --- linux-2.6.21-rc5.orig/fs/smbfs/request.c
> +++ linux-2.6.21-rc5/fs/smbfs/request.c
> @@ -347,6 +347,7 @@ int smb_add_request(struct smb_request *
>  			smb_rput(req);
>  		}
>  		smb_unlock_server(server);
> +		return -EINTR;
>  	}
>  
>  	if (!timeleft) {
> #<EOF>

This can return -EINTR if timeleft == 0 && !signal_pending(), which seems
wrong.

[PATCH 12/16] cifs-use-mutex.diff

[Jan Engelhardt]

Verbatim copy of original mail:

*>>>
>From http://lkml.org/lkml/2006/9/2/86

The recent change to "allow Windows blocking locks to be cancelled
via a CANCEL_LOCK call" introduced a new semaphore in struct
cifsFileInfo, lock_sem.  However, semaphores used as mutexes are
deprecated these days, and there's no reason to add a new one to the
kernel. Therefore, convert lock_sem to a struct mutex (and also fix
one indentation glitch on one of the lines changed anyway).

Compile tested only, since I don't use CIFS.

Signed-off-by: Roland Dreier <roland@digitalvampire.org>
*<<<


My 2¢:
Compile tested. Runtime tested. Did not break.

From: Roland Dreier <roland@digitalvampire.org>
Signed-off-by: Jan Engelhardt <jengelh@gmx.de>
Cc: Roland Dreier <roland@digitalvampire.org>

 cifsglob.h |    2 +-
 dir.c      |    2 +-
 file.c     |   14 +++++++-------
 3 files changed, 9 insertions(+), 9 deletions(-)

Index: linux-2.6.21-rc5/fs/cifs/cifsglob.h
===================================================================
--- linux-2.6.21-rc5.orig/fs/cifs/cifsglob.h
+++ linux-2.6.21-rc5/fs/cifs/cifsglob.h
@@ -311,7 +311,7 @@ struct cifsFileInfo {
 	/* lock scope id (0 if none) */
 	struct file * pfile; /* needed for writepage */
 	struct inode * pInode; /* needed for oplock break */
-	struct semaphore lock_sem;
+	struct mutex lock_mutex;
 	struct list_head llist; /* list of byte range locks we have. */
 	unsigned closePend:1;	/* file is marked to close */
 	unsigned invalidHandle:1;  /* file closed via session abend */
Index: linux-2.6.21-rc5/fs/cifs/dir.c
===================================================================
--- linux-2.6.21-rc5.orig/fs/cifs/dir.c
+++ linux-2.6.21-rc5/fs/cifs/dir.c
@@ -274,7 +274,7 @@ cifs_create(struct inode *inode, struct 
 			pCifsFile->invalidHandle = FALSE;
 			pCifsFile->closePend     = FALSE;
 			init_MUTEX(&pCifsFile->fh_sem);
-			init_MUTEX(&pCifsFile->lock_sem);
+			mutex_init(&pCifsFile->lock_mutex);
 			INIT_LIST_HEAD(&pCifsFile->llist);
 			atomic_set(&pCifsFile->wrtPending,0);
 
Index: linux-2.6.21-rc5/fs/cifs/file.c
===================================================================
--- linux-2.6.21-rc5.orig/fs/cifs/file.c
+++ linux-2.6.21-rc5/fs/cifs/file.c
@@ -48,7 +48,7 @@ static inline struct cifsFileInfo *cifs_
 	private_data->netfid = netfid;
 	private_data->pid = current->tgid;	
 	init_MUTEX(&private_data->fh_sem);
-	init_MUTEX(&private_data->lock_sem);
+	mutex_init(&private_data->lock_mutex);
 	INIT_LIST_HEAD(&private_data->llist);
 	private_data->pfile = file; /* needed for writepage */
 	private_data->pInode = inode;
@@ -508,12 +508,12 @@ int cifs_close(struct inode *inode, stru
 
 		/* Delete any outstanding lock records.
 		   We'll lose them when the file is closed anyway. */
-		down(&pSMBFile->lock_sem);
+		mutex_lock(&pSMBFile->lock_mutex);
 		list_for_each_entry_safe(li, tmp, &pSMBFile->llist, llist) {
 			list_del(&li->llist);
 			kfree(li);
 		}
-		up(&pSMBFile->lock_sem);
+		mutex_unlock(&pSMBFile->lock_mutex);
 
 		write_lock(&GlobalSMBSeslock);
 		list_del(&pSMBFile->flist);
@@ -598,9 +598,9 @@ static int store_file_lock(struct cifsFi
 	li->offset = offset;
 	li->length = len;
 	li->type = lockType;
-	down(&fid->lock_sem);
+	mutex_lock(&fid->lock_mutex);
 	list_add(&li->llist, &fid->llist);
-	up(&fid->lock_sem);
+	mutex_unlock(&fid->lock_mutex);
 	return 0;
 }
 
@@ -757,7 +757,7 @@ int cifs_lock(struct file *file, int cmd
 			struct cifsLockInfo *li, *tmp;
 
 			rc = 0;
-			down(&fid->lock_sem);
+			mutex_lock(&fid->lock_mutex);
 			list_for_each_entry_safe(li, tmp, &fid->llist, llist) {
 				if (pfLock->fl_start <= li->offset &&
 						length >= li->length) {
@@ -771,7 +771,7 @@ int cifs_lock(struct file *file, int cmd
 					kfree(li);
 				}
 			}
-			up(&fid->lock_sem);
+			mutex_unlock(&fid->lock_mutex);
 		}
 	}
 
#<EOF>

Re: [PATCH 12/16] cifs-use-mutex.diff

[Roland Dreier]

Thanks for resurrecting this.  My original email to the CIFS
maintainer got dropped somewhere, but I think this patch is probably
still a good idea.
-- 
Roland Dreier  <roland@digitalvampire.org>  GPG Key: 1024D/E0EEFAC0
Fingerprint:     A89F B5E9 C185 F34D BD50  4009 37E2 25CC E0EE FAC0

 Sending >500KB attachments is forbidden by the Geneva Convention.
        Your country may be at risk if you fail to comply.

[PATCH 13/16] show-pipesize-in-stat.diff

[Jan Engelhardt]

Show the fill status of a pipe (in bytes) when stat'ing one.

Signed-off-by: Jan Engelhardt <jengelh@gmx.de>

 fs/stat.c             |   31 ++++++++++++++++++++++++++++++-
 include/linux/un.h    |    2 ++
 include/net/af_unix.h |    3 +++
 net/unix/af_unix.c    |   10 ++++++++++
 4 files changed, 45 insertions(+), 1 deletion(-)

Index: linux-2.6.21-rc5/fs/stat.c
===================================================================
--- linux-2.6.21-rc5.orig/fs/stat.c
+++ linux-2.6.21-rc5/fs/stat.c
@@ -14,6 +14,10 @@
 #include <linux/namei.h>
 #include <linux/security.h>
 #include <linux/syscalls.h>
+#include <linux/blkdev.h>
+#include <linux/genhd.h>
+#include <linux/pipe_fs_i.h>
+#include <net/af_unix.h>
 #include <linux/pagemap.h>
 
 #include <asm/uaccess.h>
@@ -31,7 +35,32 @@ void generic_fillattr(struct inode *inod
 	stat->atime = inode->i_atime;
 	stat->mtime = inode->i_mtime;
 	stat->ctime = inode->i_ctime;
-	stat->size = i_size_read(inode);
+	stat->size = 0;
+	if (S_ISFIFO(inode->i_mode)) {
+		const struct pipe_inode_info *info = inode->i_pipe;
+		int i;
+		stat->size = 0;
+		if (info != NULL) {
+			for (i = 0; i < PIPE_BUFFERS; ++i) {
+				const struct pipe_buffer *buf = &info->bufs[i];
+				if (buf != NULL && buf->page != NULL)
+					stat->size += buf->len;
+			}
+		}
+	} else if (S_ISSOCK(inode->i_mode)) {
+#ifdef CONFIG_UNIX_MODULE
+		loff_t (*uxsize)(struct inode *) = __symbol_get("unixsock_size");
+		if (uxsize != NULL) {
+			stat->size = uxsize(inode);
+			symbol_put("unixsock_size");
+		}
+#endif
+#if defined(CONFIG_UNIX) && !defined(CONFIG_UNIX_MODULE)
+		stat->size = unixsock_size(inode);
+#endif
+	} else {
+		stat->size = i_size_read(inode);
+	}
 	stat->blocks = inode->i_blocks;
 	stat->blksize = (1 << inode->i_blkbits);
 }
Index: linux-2.6.21-rc5/include/linux/un.h
===================================================================
--- linux-2.6.21-rc5.orig/include/linux/un.h
+++ linux-2.6.21-rc5/include/linux/un.h
@@ -3,6 +3,8 @@
 
 #define UNIX_PATH_MAX	108
 
+#include <linux/socket.h>
+
 struct sockaddr_un {
 	sa_family_t sun_family;	/* AF_UNIX */
 	char sun_path[UNIX_PATH_MAX];	/* pathname */
Index: linux-2.6.21-rc5/include/net/af_unix.h
===================================================================
--- linux-2.6.21-rc5.orig/include/net/af_unix.h
+++ linux-2.6.21-rc5/include/net/af_unix.h
@@ -2,6 +2,7 @@
 #define __LINUX_NET_AFUNIX_H
 
 #include <linux/socket.h>
+#include <linux/types.h>
 #include <linux/un.h>
 #include <linux/mutex.h>
 #include <net/sock.h>
@@ -88,6 +89,8 @@ struct unix_sock {
 };
 #define unix_sk(__sk) ((struct unix_sock *)__sk)
 
+extern loff_t unixsock_size(struct inode *);
+
 #ifdef CONFIG_SYSCTL
 extern int sysctl_unix_max_dgram_qlen;
 extern void unix_sysctl_register(void);
Index: linux-2.6.21-rc5/net/unix/af_unix.c
===================================================================
--- linux-2.6.21-rc5.orig/net/unix/af_unix.c
+++ linux-2.6.21-rc5/net/unix/af_unix.c
@@ -744,6 +744,16 @@ fail:
 	return NULL;
 }
 
+loff_t unixsock_size(struct inode *inode) {
+    struct sock *sk = unix_find_socket_byinode(inode);
+    loff_t eax;
+
+    if(sk == NULL) { return 0; }
+    eax = sk->sk_rcvbuf;
+    sock_put(sk);
+    return eax;
+}
+EXPORT_SYMBOL(unixsock_size);
 
 static int unix_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len)
 {
#<EOF>

Re: [PATCH 13/16] show-pipesize-in-stat.diff

[Andrew Morton]

On Sun, 1 Apr 2007 20:17:24 +0200 (MEST) Jan Engelhardt <jengelh@linux01.gwdg.de> wrote:

> 
> Show the fill status of a pipe (in bytes) when stat'ing one.
> 

Is this useful?  It seems rather an obscure thing, and we generally need a
good reason to go adding Linux-specific goodies to standard system calls
like this.  What do "other operating systems" do here?

> -	stat->size = i_size_read(inode);
> +	stat->size = 0;
> +	if (S_ISFIFO(inode->i_mode)) {
> +		const struct pipe_inode_info *info = inode->i_pipe;
> +		int i;
> +		stat->size = 0;
> +		if (info != NULL) {
> +			for (i = 0; i < PIPE_BUFFERS; ++i) {
> +				const struct pipe_buffer *buf = &info->bufs[i];
> +				if (buf != NULL && buf->page != NULL)
> +					stat->size += buf->len;
> +			}
> +		}
> +	} else if (S_ISSOCK(inode->i_mode)) {
> +#ifdef CONFIG_UNIX_MODULE
> +		loff_t (*uxsize)(struct inode *) = __symbol_get("unixsock_size");
> +		if (uxsize != NULL) {
> +			stat->size = uxsize(inode);
> +			symbol_put("unixsock_size");
> +		}
> +#endif
> +#if defined(CONFIG_UNIX) && !defined(CONFIG_UNIX_MODULE)
> +		stat->size = unixsock_size(inode);
> +#endif
> +	} else {
> +		stat->size = i_size_read(inode);
> +	}

That's a bit fugly.

> +loff_t unixsock_size(struct inode *inode) {
> +    struct sock *sk = unix_find_socket_byinode(inode);
> +    loff_t eax;
> +
> +    if(sk == NULL) { return 0; }
> +    eax = sk->sk_rcvbuf;
> +    sock_put(sk);
> +    return eax;
> +}
> +EXPORT_SYMBOL(unixsock_size);

That's wildly incorrect coding style, and it doesn't use tabs.

Re: [PATCH 13/16] show-pipesize-in-stat.diff

[Jan Engelhardt]

On Apr 1 2007 22:58, Andrew Morton wrote:
>On Sun, 1 Apr 2007 20:17:24 +0200 (MEST) Jan Engelhardt <jengelh@linux01.gwdg.de> wrote:
>
>> Show the fill status of a pipe (in bytes) when stat'ing one.
>
>Is this useful?  It seems rather an obscure thing, and we generally need a
>good reason to go adding Linux-specific goodies to standard system calls
>like this.  What do "other operating systems" do here?

Other OS seem to always show 0, as Linux does usually. So I do not think
programs actually rely on it having any specific value.
I thought adding pipesize might be helpful in dianosing pipe cases, akin to
Miklos's procfs-fdinfo patch for regular files.

>> +		}
>> +	} else if (S_ISSOCK(inode->i_mode)) {
>> +#ifdef CONFIG_UNIX_MODULE
>> +		loff_t (*uxsize)(struct inode *) = __symbol_get("unixsock_size");
>> +		if (uxsize != NULL) {
>> +			stat->size = uxsize(inode);
>> +			symbol_put("unixsock_size");
>> +		}
>> +#endif
>> +#if defined(CONFIG_UNIX) && !defined(CONFIG_UNIX_MODULE)
>> +		stat->size = unixsock_size(inode);
>> +#endif
>> +	} else {
>> +		stat->size = i_size_read(inode);
>> +	}
>
>That's a bit fugly.

Drop for now.

>> +loff_t unixsock_size(struct inode *inode) {
>> +    struct sock *sk = unix_find_socket_byinode(inode);
>> +    loff_t eax;
>> +
>> +    if(sk == NULL) { return 0; }
>> +    eax = sk->sk_rcvbuf;
>> +    sock_put(sk);
>> +    return eax;
>> +}
>> +EXPORT_SYMBOL(unixsock_size);
>
>That's wildly incorrect coding style, and it doesn't use tabs.

Which means I something really went wrong, because I knew I went through it for
cleanup.

Re: [PATCH 13/16] show-pipesize-in-stat.diff

[Eric Dumazet]

On Sun, 1 Apr 2007 20:17:24 +0200 (MEST)
Jan Engelhardt <jengelh@linux01.gwdg.de> wrote:

> 
> Show the fill status of a pipe (in bytes) when stat'ing one.
> 
> Signed-off-by: Jan Engelhardt <jengelh@gmx.de>
> 
>  fs/stat.c             |   31 ++++++++++++++++++++++++++++++-
>  include/linux/un.h    |    2 ++
>  include/net/af_unix.h |    3 +++
>  net/unix/af_unix.c    |   10 ++++++++++
>  4 files changed, 45 insertions(+), 1 deletion(-)

Please dont do that, adding uggly tests in generic_fillattr()

The following patch does the right thing for pipes, I let you doing the same for sockets...

[PATCH] : fill the size of pipes

Instead of reporting 0 in size when stating() a pipe, we give the number of queued bytes. This might avoid using ioctl(FIONREAD) to get this information.

Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
---
--- linux-2.6.21-rc5-mm3/fs/pipe.c
+++ linux-2.6.21-rc5-mm3/fs/pipe.c
@@ -570,27 +570,34 @@ bad_pipe_w(struct file *filp, const char
 	return -EBADF;
 }
 
+static int pipe_size(struct inode *inode)
+{
+	struct pipe_inode_info *pipe;
+	int count, buf, nrbufs;
+
+	mutex_lock(&inode->i_mutex);
+	pipe = inode->i_pipe;
+	count = 0;
+	buf = pipe->curbuf;
+	nrbufs = pipe->nrbufs;
+	while (--nrbufs >= 0) {
+		count += pipe->bufs[buf].len;
+		buf = (buf+1) & (PIPE_BUFFERS-1);
+	}
+	mutex_unlock(&inode->i_mutex);
+	return count;
+}
+
 static int
 pipe_ioctl(struct inode *pino, struct file *filp,
 	   unsigned int cmd, unsigned long arg)
 {
 	struct inode *inode = filp->f_path.dentry->d_inode;
-	struct pipe_inode_info *pipe;
-	int count, buf, nrbufs;
+	int count;
 
 	switch (cmd) {
 		case FIONREAD:
-			mutex_lock(&inode->i_mutex);
-			pipe = inode->i_pipe;
-			count = 0;
-			buf = pipe->curbuf;
-			nrbufs = pipe->nrbufs;
-			while (--nrbufs >= 0) {
-				count += pipe->bufs[buf].len;
-				buf = (buf+1) & (PIPE_BUFFERS-1);
-			}
-			mutex_unlock(&inode->i_mutex);
-
+			count = pipe_size(inode);
 			return put_user(count, (int __user *)arg);
 		default:
 			return -EINVAL;
@@ -908,6 +915,20 @@ static struct dentry_operations pipefs_d
 	.d_dname	= pipefs_dname,
 };
 
+int pipe_getattr(struct vfsmount *mnt, struct dentry *dentry,
+		   struct kstat *stat)
+{
+	struct inode *inode = dentry->d_inode;
+
+	generic_fillattr(inode, stat);
+	stat->size = pipe_size(inode);
+	return 0;
+}
+
+const struct inode_operations pipe_inode_operations = {
+	.getattr = pipe_getattr,
+};
+
 static struct inode * get_pipe_inode(void)
 {
 	struct inode *inode = new_inode(pipe_mnt->mnt_sb);
@@ -921,6 +942,7 @@ static struct inode * get_pipe_inode(voi
 		goto fail_iput;
 	inode->i_pipe = pipe;
 
+	inode->i_op = &pipe_inode_operations;
 	pipe->readers = pipe->writers = 1;
 	inode->i_fop = &rdwr_pipe_fops;
 

Re: [PATCH 13/16] show-pipesize-in-stat.diff

[Jan Engelhardt]

On Apr 2 2007 12:41, Eric Dumazet wrote:
>
>The following patch does the right thing for pipes, I let you doing
>the same for sockets...

Is struct sock->sk_receive_queue->qlen actually the thing I am
looking for? (Previously, I had struct sock->sk_rcvbuf, which was
obviously wrong.)

Thanks,
Jan
-- 

Re: [PATCH 13/16] show-pipesize-in-stat.diff

[Eric Dumazet]

Jan Engelhardt a écrit :
> On Apr 2 2007 12:41, Eric Dumazet wrote:
>> The following patch does the right thing for pipes, I let you doing
>> the same for sockets...
> 
> Is struct sock->sk_receive_queue->qlen actually the thing I am
> looking for? (Previously, I had struct sock->sk_rcvbuf, which was
> obviously wrong.)
> 

find net | xargs grep -n SIOCINQ

I suspect you will find unix_ioctl() in net/unix/af_unix.c ?

BTW, my patch only worked for pipes, not fifos. Ie pipe(), not 
open("/some/fifo.p", ...)

Re: [PATCH 13/16] show-pipesize-in-stat.diff

[Jan Engelhardt]

Hi,

On Wed, 04 Apr 2007 07:03:55 +0200, Eric Dumazet wrote:
>find net | xargs grep -n SIOCINQ
>I suspect you will find unix_ioctl() in net/unix/af_unix.c ?
>BTW, my patch only worked for pipes, not fifos. Ie pipe(), not 
>open("/some/fifo.p", ...)

Thanks for your patch. There is a question however; why don't you do
  stat->size = pipe_size(inode);
in generic_fill_attr (perhaps wrapped into an if(S_ISFIFO))?


Regards,
Jan
-- 

[PATCH 14/16] kconfig-allow-override.diff

[Jan Engelhardt]

Allow config variables in .config to override earlier ones in the same
file. In other words,

	# CONFIG_SECURITY is not defined
	CONFIG_SECURITY=y

will activate it. This makes it a bit easier to do

	(cat original-config myconfig myconfig2 ... >.config)

and run menuconfig as expected.

Already posted at http://lkml.org/lkml/2006/10/25/81
Signed-off-by: Jan Engelhardt <jengelh@gmx.de>
Cc: Sam Ravnborg <sam@ravnborg.org>

 confdata.c |   14 +++++---------
 1 file changed, 5 insertions(+), 9 deletions(-)

Index: linux-2.6.21-rc5/scripts/kconfig/confdata.c
===================================================================
--- linux-2.6.21-rc5.orig/scripts/kconfig/confdata.c
+++ linux-2.6.21-rc5/scripts/kconfig/confdata.c
@@ -170,8 +170,7 @@ load:
 					sym->type = S_BOOLEAN;
 			}
 			if (sym->flags & def_flags) {
-				conf_warning("trying to reassign symbol %s", sym->name);
-				break;
+				conf_warning("override: reassigning to symbol %s", sym->name);
 			}
 			switch (sym->type) {
 			case S_BOOLEAN:
@@ -210,8 +209,7 @@ load:
 					sym->type = S_OTHER;
 			}
 			if (sym->flags & def_flags) {
-				conf_warning("trying to reassign symbol %s", sym->name);
-				break;
+				conf_warning("override: reassigning to symbol %s", sym->name);
 			}
 			switch (sym->type) {
 			case S_TRISTATE:
@@ -288,11 +286,9 @@ load:
 				}
 				break;
 			case yes:
-				if (cs->def[def].tri != no) {
-					conf_warning("%s creates inconsistent choice state", sym->name);
-					cs->flags &= ~def_flags;
-				} else
-					cs->def[def].val = sym;
+				if(cs->def[def].tri != no)
+					conf_warning("override: %s turns state choice", sym->name);
+				cs->def[def].val = sym;
 				break;
 			}
 			cs->def[def].tri = E_OR(cs->def[def].tri, sym->def[def].tri);
#<EOF>

Re: [PATCH 14/16] kconfig-allow-override.diff

[Sam Ravnborg]

This part of kconfig is taken care of by Roman Zippel.
Kept the full mail for his reference.

	Sam

On Sun, Apr 01, 2007 at 08:17:32PM +0200, Jan Engelhardt wrote:
> 
> Allow config variables in .config to override earlier ones in the same
> file. In other words,
> 
> 	# CONFIG_SECURITY is not defined
> 	CONFIG_SECURITY=y
> 
> will activate it. This makes it a bit easier to do
> 
> 	(cat original-config myconfig myconfig2 ... >.config)
> 
> and run menuconfig as expected.
> 
> Already posted at http://lkml.org/lkml/2006/10/25/81
> Signed-off-by: Jan Engelhardt <jengelh@gmx.de>
> Cc: Sam Ravnborg <sam@ravnborg.org>
> 
>  confdata.c |   14 +++++---------
>  1 file changed, 5 insertions(+), 9 deletions(-)
> 
> Index: linux-2.6.21-rc5/scripts/kconfig/confdata.c
> ===================================================================
> --- linux-2.6.21-rc5.orig/scripts/kconfig/confdata.c
> +++ linux-2.6.21-rc5/scripts/kconfig/confdata.c
> @@ -170,8 +170,7 @@ load:
>  					sym->type = S_BOOLEAN;
>  			}
>  			if (sym->flags & def_flags) {
> -				conf_warning("trying to reassign symbol %s", sym->name);
> -				break;
> +				conf_warning("override: reassigning to symbol %s", sym->name);
>  			}
>  			switch (sym->type) {
>  			case S_BOOLEAN:
> @@ -210,8 +209,7 @@ load:
>  					sym->type = S_OTHER;
>  			}
>  			if (sym->flags & def_flags) {
> -				conf_warning("trying to reassign symbol %s", sym->name);
> -				break;
> +				conf_warning("override: reassigning to symbol %s", sym->name);
>  			}
>  			switch (sym->type) {
>  			case S_TRISTATE:
> @@ -288,11 +286,9 @@ load:
>  				}
>  				break;
>  			case yes:
> -				if (cs->def[def].tri != no) {
> -					conf_warning("%s creates inconsistent choice state", sym->name);
> -					cs->flags &= ~def_flags;
> -				} else
> -					cs->def[def].val = sym;
> +				if(cs->def[def].tri != no)
> +					conf_warning("override: %s turns state choice", sym->name);
> +				cs->def[def].val = sym;
>  				break;
>  			}
>  			cs->def[def].tri = E_OR(cs->def[def].tri, sym->def[def].tri);
> #<EOF>

Re: [PATCH 14/16] kconfig-allow-override.diff

[Randy Dunlap]

On Sun, 1 Apr 2007 20:44:41 +0200 Sam Ravnborg wrote:

> This part of kconfig is taken care of by Roman Zippel.
> Kept the full mail for his reference.
> 
> 	Sam

I certainly want this.


> On Sun, Apr 01, 2007 at 08:17:32PM +0200, Jan Engelhardt wrote:
> > 
> > Allow config variables in .config to override earlier ones in the same
> > file. In other words,
> > 
> > 	# CONFIG_SECURITY is not defined
> > 	CONFIG_SECURITY=y
> > 
> > will activate it. This makes it a bit easier to do
> > 
> > 	(cat original-config myconfig myconfig2 ... >.config)
> > 
> > and run menuconfig as expected.
> > 
> > Already posted at http://lkml.org/lkml/2006/10/25/81
> > Signed-off-by: Jan Engelhardt <jengelh@gmx.de>
> > Cc: Sam Ravnborg <sam@ravnborg.org>
> > 
> >  confdata.c |   14 +++++---------
> >  1 file changed, 5 insertions(+), 9 deletions(-)
> > 
> > Index: linux-2.6.21-rc5/scripts/kconfig/confdata.c
> > ===================================================================
> > --- linux-2.6.21-rc5.orig/scripts/kconfig/confdata.c
> > +++ linux-2.6.21-rc5/scripts/kconfig/confdata.c
> > @@ -170,8 +170,7 @@ load:
> >  					sym->type = S_BOOLEAN;
> >  			}
> >  			if (sym->flags & def_flags) {
> > -				conf_warning("trying to reassign symbol %s", sym->name);
> > -				break;
> > +				conf_warning("override: reassigning to symbol %s", sym->name);
> >  			}
> >  			switch (sym->type) {
> >  			case S_BOOLEAN:
> > @@ -210,8 +209,7 @@ load:
> >  					sym->type = S_OTHER;
> >  			}
> >  			if (sym->flags & def_flags) {
> > -				conf_warning("trying to reassign symbol %s", sym->name);
> > -				break;
> > +				conf_warning("override: reassigning to symbol %s", sym->name);
> >  			}
> >  			switch (sym->type) {
> >  			case S_TRISTATE:
> > @@ -288,11 +286,9 @@ load:
> >  				}
> >  				break;
> >  			case yes:
> > -				if (cs->def[def].tri != no) {
> > -					conf_warning("%s creates inconsistent choice state", sym->name);
> > -					cs->flags &= ~def_flags;
> > -				} else
> > -					cs->def[def].val = sym;
> > +				if(cs->def[def].tri != no)
> > +					conf_warning("override: %s turns state choice", sym->name);
> > +				cs->def[def].val = sym;
> >  				break;
> >  			}
> >  			cs->def[def].tri = E_OR(cs->def[def].tri, sym->def[def].tri);
> > #<EOF>


---
~Randy
*** Remember to use Documentation/SubmitChecklist when testing your code ***

[PATCH 15/16] use-regular-eth-suffix.diff

[Jan Engelhardt]

Some radio adapter drivers wrongly(?) name their devices "wlan%d"
instead of "eth%d" (if you ask me, it should be %u - but not today).

Technically, they operate like Ethernet, and in fact, running
`/sbin/ip a` shows "link/ether" instead of "link/ieee80211".
This patch renames them back, but I would appreciate some comment,
explanation or at least link why they actually have wlan%d there.

Signed-off-by: Jan Engelhardt <jengelh@gmx.de>
Cc: Jouni Malinen <jkmaline@cc.hut.fi> (hostap)
Cc: Jeroen Vreeken <pe1rxq@amsat.org> (zd1201)

 hostap/hostap_hw.c |    4 ++--
 zd1201.c           |    2 +-
 2 files changed, 3 insertions(+), 3 deletions(-)

Index: linux-2.6.21-rc5/drivers/net/wireless/hostap/hostap_hw.c
===================================================================
--- linux-2.6.21-rc5.orig/drivers/net/wireless/hostap/hostap_hw.c
+++ linux-2.6.21-rc5/drivers/net/wireless/hostap/hostap_hw.c
@@ -81,10 +81,10 @@ static int dtim_period[MAX_PARM_DEVICES]
 module_param_array(dtim_period, int, NULL, 0444);
 MODULE_PARM_DESC(dtim_period, "DTIM period");
 
-static char dev_template[16] = "wlan%d";
+static char dev_template[16] = "eth%d";
 module_param_string(dev_template, dev_template, sizeof(dev_template), 0444);
 MODULE_PARM_DESC(dev_template, "Prefix for network device name (default: "
-		 "wlan%d)");
+		 "eth%d)");
 
 #ifdef final_version
 #define EXTRA_EVENTS_WTERR 0
Index: linux-2.6.21-rc5/drivers/net/wireless/zd1201.c
===================================================================
--- linux-2.6.21-rc5.orig/drivers/net/wireless/zd1201.c
+++ linux-2.6.21-rc5/drivers/net/wireless/zd1201.c
@@ -1787,7 +1787,7 @@ static int zd1201_probe(struct usb_inter
 	zd->dev->tx_timeout = zd1201_tx_timeout;
 	zd->dev->set_multicast_list = zd1201_set_multicast;
 	zd->dev->set_mac_address = zd1201_set_mac_address;
-	strcpy(zd->dev->name, "wlan%d");
+	strcpy(zd->dev->name, "eth%d");
 
 	err = zd1201_getconfig(zd, ZD1201_RID_CNFOWNMACADDR, 
 	    zd->dev->dev_addr, zd->dev->addr_len);
#<EOF>

Re: [PATCH 15/16] use-regular-eth-suffix.diff

[Kyle Moffett]

On Apr 01, 2007, at 14:18:12, Jan Engelhardt wrote:
> Some radio adapter drivers wrongly(?) name their devices "wlan%d"  
> instead of "eth%d" (if you ask me, it should be %u - but not today).
>
> Technically, they operate like Ethernet, and in fact, running `/ 
> sbin/ip a` shows "link/ether" instead of "link/ieee80211". This  
> patch renames them back, but I would appreciate some comment,  
> explanation or at least link why they actually have wlan%d there.

NACK.  This has the potential to significantly break backwards- 
compatibility for users of those cards who are not using udev to  
maintain consistent device naming (because they only have a single  
card, for example).  I would recommend against applying it for that  
reason.

Cheers,
Kyle Moffett

Re: [PATCH 15/16] use-regular-eth-suffix.diff

[Jouni Malinen]

On Sun, Apr 01, 2007 at 08:18:12PM +0200, Jan Engelhardt wrote:

> Some radio adapter drivers wrongly(?) name their devices "wlan%d"
> instead of "eth%d" (if you ask me, it should be %u - but not today).

> Technically, they operate like Ethernet, and in fact, running
> `/sbin/ip a` shows "link/ether" instead of "link/ieee80211".

The address may look like wired Ethernet (both are based on IEEE 802
standards), but IEEE 802.11 does not behave like IEEE 802.3. For
example, one cannot send frames with a foreign source address in client
mode when using IEEE 802.11 which is clearly different from wired
Ethernet. IEEE 802.11 interfaces (in client mode) cannot be bridged at
layer 2 in the same way as wired Ethernet interfaces could.

> This patch renames them back, but I would appreciate some comment,
> explanation or at least link why they actually have wlan%d there.

Renames _back_? No it doesn't. Host AP driver has never used eth%d as
the interface name template

> Index: linux-2.6.21-rc5/drivers/net/wireless/hostap/hostap_hw.c
> -static char dev_template[16] = "wlan%d";
> +static char dev_template[16] = "eth%d";

NAK. Host AP driver has been using wlan%d for close to seven years and I
see no reason to change it now. This will just cause problems for users.
If someone wants to rename the interface to something else, they can do
it with number of different ways from user space. The kernel default
should not be changed at this point.

-- 
Jouni Malinen                                            PGP id EFC895FA

[PATCH 16/16] warn-on-kthread-name-truncation.diff

[Jan Engelhardt]

Andrew Morton noted in http://lkml.org/lkml/2006/6/30/247

	"""We do occasionally hit task_struct.comm[] truncation, when people
	use "too-long-a-name%d" for their kernel thread names."""

This patch warns when such a truncation happens.

Already posted on http://lkml.org/lkml/2006/7/3/93

Signed-off-by: Jan Engelhardt <jengelh@gmx.de>
Cc: Andrew Morton <akpm@linux-foundation.org>

 kthread.c |    7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

Index: linux-2.6.21-rc5/kernel/kthread.c
===================================================================
--- linux-2.6.21-rc5.orig/kernel/kthread.c
+++ linux-2.6.21-rc5/kernel/kthread.c
@@ -176,8 +176,11 @@ struct task_struct *kthread_create(int (
 	if (!IS_ERR(create.result)) {
 		va_list args;
 		va_start(args, namefmt);
-		vsnprintf(create.result->comm, sizeof(create.result->comm),
-			  namefmt, args);
+		if(vsnprintf(create.result->comm, sizeof(create.result->comm),
+		  namefmt, args) != strlen(create.result->comm))
+			printk(KERN_WARNING "kthread_create: command name of "
+			  "pid %d truncated to \"%s\"\n", create.result->pid,
+			  create.result->comm);
 		va_end(args);
 	}
 
#<EOF>

Re: [PATCH 16/16] warn-on-kthread-name-truncation.diff

[Andrew Morton]

On Sun, 1 Apr 2007 20:18:22 +0200 (MEST) Jan Engelhardt <jengelh@linux01.gwdg.de> wrote:

> Andrew Morton noted in http://lkml.org/lkml/2006/6/30/247
> 
> 	"""We do occasionally hit task_struct.comm[] truncation, when people
> 	use "too-long-a-name%d" for their kernel thread names."""
> 
> This patch warns when such a truncation happens.
> 
> Already posted on http://lkml.org/lkml/2006/7/3/93
> 
> Signed-off-by: Jan Engelhardt <jengelh@gmx.de>
> Cc: Andrew Morton <akpm@linux-foundation.org>
> 
>  kthread.c |    7 +++++--
>  1 file changed, 5 insertions(+), 2 deletions(-)
> 
> Index: linux-2.6.21-rc5/kernel/kthread.c
> ===================================================================
> --- linux-2.6.21-rc5.orig/kernel/kthread.c
> +++ linux-2.6.21-rc5/kernel/kthread.c
> @@ -176,8 +176,11 @@ struct task_struct *kthread_create(int (
>  	if (!IS_ERR(create.result)) {
>  		va_list args;
>  		va_start(args, namefmt);
> -		vsnprintf(create.result->comm, sizeof(create.result->comm),
> -			  namefmt, args);
> +		if(vsnprintf(create.result->comm, sizeof(create.result->comm),
> +		  namefmt, args) != strlen(create.result->comm))
> +			printk(KERN_WARNING "kthread_create: command name of "
> +			  "pid %d truncated to \"%s\"\n", create.result->pid,
> +			  create.result->comm);
>  		va_end(args);
>  	}

I'm not sure that it's a big enough problem to go adding code for, really.

Re: [PATCH 16/16] warn-on-kthread-name-truncation.diff

[Jan Engelhardt]

On Apr 1 2007 23:00, Andrew Morton wrote:
>On Sun, 1 Apr 2007 20:18:22 +0200 (MEST) Jan Engelhardt <jengelh@linux01.gwdg.de> wrote:
>
>I'm not sure that it's a big enough problem to go adding code for, really.

No, not really. If there are any kthread spawners that do exceed the limit,
I would have probably gotten a message by now over the last handful of kernel
releases. So everything seems clear.


Jan
--