idebus setup problem (2.6.7-rc1)

idebus setup problem (2.6.7-rc1)

[Auzanneau Gregory]

[-- Attachment #1: Type: text/plain, Size: 745 bytes --]


Hello,

It seems there is a problem with idebus parameter with 2.6.7-rc1.
Indeed, it doesn't take into account lilo append.

With 2.6.7-rc1-mm1, i've got:
Kernel command line: BOOT_IMAGE=LinuxNEW ro root=304 idebus=66
ide: Assuming 33MHz system bus speed for PIO modes; override with idebus=xx

With 2.6.6-mm3, i've got:
May 24 11:37:43 greg-port kernel: Kernel command line:
BOOT_IMAGE=LinuxNEW ro root=304 idebus=66
May 24 11:37:43 greg-port kernel: ide_setup: idebus=66
May 24 11:37:43 greg-port kernel: ide: Assuming 66MHz system bus speed
for PIO modes

I tried to seek in the code, but my level is not as good as I would like
it. :)


Thank you all for the good work with linux, keep up with it ! :)

-- 
Auzanneau Grégory
GPG 0x99137BEE

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 251 bytes --]

Re: idebus setup problem (2.6.7-rc1)

[Bartlomiej Zolnierkiewicz]

On Thursday 27 of May 2004 13:57, Auzanneau Gregory wrote:
> Hello,
>
> It seems there is a problem with idebus parameter with 2.6.7-rc1.
> Indeed, it doesn't take into account lilo append.

Why are you using it in the first place?

Let me guess... you've nForce2 chipset and lspci shows you 66MHz.
You don't need and really shouldn't be using 'idebus=66'.

> With 2.6.7-rc1-mm1, i've got:
> Kernel command line: BOOT_IMAGE=LinuxNEW ro root=304 idebus=66
> ide: Assuming 33MHz system bus speed for PIO modes; override with idebus=xx

This needs fixing.

I remember seeing patch related to handling '=' in kernel params,
maybe it's related (or maybe not).

> With 2.6.6-mm3, i've got:
> May 24 11:37:43 greg-port kernel: Kernel command line:
> BOOT_IMAGE=LinuxNEW ro root=304 idebus=66
> May 24 11:37:43 greg-port kernel: ide_setup: idebus=66
> May 24 11:37:43 greg-port kernel: ide: Assuming 66MHz system bus speed
> for PIO modes
>
> I tried to seek in the code, but my level is not as good as I would like
> it. :)
>
>
> Thank you all for the good work with linux, keep up with it ! :)

RE: idebus setup problem (2.6.7-rc1)

[Zhu, Yi]

Bartlomiej Zolnierkiewicz [B.Zolnierkiewicz@elka.pw.edu.pl] wrote:
> 
> I remember seeing patch related to handling '=' in kernel
> params, maybe it's related (or maybe not).

Yes, this is caused by my kernel-parameter-parsing-fix.patch.

But I think below code in ide.c is a hack.
__setup("", ide_setup);

How about below change?

--- linux-2.6.7-rc1-mm1.orig/drivers/ide/ide.c      2004-05-27
23:07:59.405138992 +0800
+++ linux-2.6.7-rc1-mm1/drivers/ide/ide.c   2004-05-27
23:09:47.529701560 +0800
@@ -2459,7 +2459,8 @@ void cleanup_module (void)

 #else /* !MODULE */

-__setup("", ide_setup);
+__setup("hd", ide_setup);
+__setup("ide", ide_setup);

 module_init(ide_init);


-yi

Re: idebus setup problem (2.6.7-rc1)

[Bartlomiej Zolnierkiewicz]

On Thursday 27 of May 2004 17:21, Zhu, Yi wrote:
> Bartlomiej Zolnierkiewicz [B.Zolnierkiewicz@elka.pw.edu.pl] wrote:
> > I remember seeing patch related to handling '=' in kernel
> > params, maybe it's related (or maybe not).
>
> Yes, this is caused by my kernel-parameter-parsing-fix.patch.
>
> But I think below code in ide.c is a hack.
> __setup("", ide_setup);

It is a really bad hack but unfortunately it is not easy to fix.
We can only do it by introducing new kernel parameters and obsoleting
current braindamage (+ halting boot if obsolete parameter is detected).

Please note that in (very) rare corner cases current breakage can cause
data corruption.

> How about below change?

It breaks all "idex=" and "hdx=" options.
Please take a look at how ide_setup().

> --- linux-2.6.7-rc1-mm1.orig/drivers/ide/ide.c      2004-05-27
> 23:07:59.405138992 +0800
> +++ linux-2.6.7-rc1-mm1/drivers/ide/ide.c   2004-05-27
> 23:09:47.529701560 +0800
> @@ -2459,7 +2459,8 @@ void cleanup_module (void)
>
>  #else /* !MODULE */
>
> -__setup("", ide_setup);
> +__setup("hd", ide_setup);
> +__setup("ide", ide_setup);
>
>  module_init(ide_init);
>
>
> -yi

Re: idebus setup problem (2.6.7-rc1)

[Jeff Garzik]

Zhu, Yi wrote:
> --- linux-2.6.7-rc1-mm1.orig/drivers/ide/ide.c      2004-05-27
> 23:07:59.405138992 +0800
> +++ linux-2.6.7-rc1-mm1/drivers/ide/ide.c   2004-05-27
> 23:09:47.529701560 +0800
> @@ -2459,7 +2459,8 @@ void cleanup_module (void)
> 
>  #else /* !MODULE */
> 
> -__setup("", ide_setup);
> +__setup("hd", ide_setup);
> +__setup("ide", ide_setup);
> 
>  module_init(ide_init);


module_param() works for both the built-in case (where __setup is used), 
and also the modular case.  If this is getting changed, might as well do 
the right change...

	Jeff

RE: idebus setup problem (2.6.7-rc1)

[Zhu, Yi]

Bartlomiej Zolnierkiewicz wrote:
> 
> It breaks all "idex=" and "hdx=" options.
> Please take a look at how ide_setup().

Yes, thanks for pointing out. Maybe we need some
wildcard support. If module_param() can do this, that's
great.

-yi

RE: idebus setup problem (2.6.7-rc1)

[Zhu, Yi]

linux-kernel-owner@vger.kernel.org wrote:
> Bartlomiej Zolnierkiewicz wrote:
>> 
>> It breaks all "idex=" and "hdx=" options.
>> Please take a look at how ide_setup().
> 
> Yes, thanks for pointing out. Maybe we need some wildcard
> support. If module_param() can do this, that's great.

Does below change acceptable to make module_param support 
wildcard '?' ?

--- linux-2.6.6.orig/kernel/params.c        2004-05-28
00:13:24.931368296 +0800
+++ linux-2.6.6/kernel/params.c     2004-05-28 00:16:04.406124448 +0800
@@ -37,7 +37,8 @@ static inline int dash2underscore(char c
 static inline int parameq(const char *input, const char *paramname)
 {
        unsigned int i;
-       for (i = 0; dash2underscore(input[i]) == paramname[i]; i++)
+       for (i = 0; paramname[i] == '?' ||
+            dash2underscore(input[i]) == paramname[i]; i++)
                if (input[i] == '\0')
                        return 1;
        return 0;

Re: idebus setup problem (2.6.7-rc1)

[Zhu, Yi]

On Thu, 27 May 2004, Auzanneau Gregory wrote:
> 
> Hello,
> 
> It seems there is a problem with idebus parameter with 2.6.7-rc1.
> Indeed, it doesn't take into account lilo append.
> 
> With 2.6.7-rc1-mm1, i've got:
> Kernel command line: BOOT_IMAGE=LinuxNEW ro root=304 idebus=66
> ide: Assuming 33MHz system bus speed for PIO modes; override with idebus=xx
> 
> With 2.6.6-mm3, i've got:
> May 24 11:37:43 greg-port kernel: Kernel command line:
> BOOT_IMAGE=LinuxNEW ro root=304 idebus=66
> May 24 11:37:43 greg-port kernel: ide_setup: idebus=66
> May 24 11:37:43 greg-port kernel: ide: Assuming 66MHz system bus speed
> for PIO modes

My bad. I missed ide_setup(), please try below patch.

--- linux-2.6.7-rc1-mm1.orig/init/main.c	2004-05-28 12:39:15.549314064 +0800
+++ linux-2.6.7-rc1-mm1/init/main.c	2004-05-28 12:40:29.399087192 +0800
@@ -162,7 +162,7 @@ static int __init obsolete_checksetup(ch
 	p = &__setup_start;
 	do {
 		int n = strlen(p->str);
-		if (len <= n && !strncmp(line, p->str, n)) {
+		if (n == 0 || (len <= n && !strncmp(line, p->str, n))) {
 			/* Already done in parse_early_param? */
 			if (p->early)
 				return 1;


Andrew, could you apply?

Thanks,
-yi

Re: idebus setup problem (2.6.7-rc1)

[Auzanneau Gregory]

Zhu, Yi a écrit :
> My bad. I missed ide_setup(), please try below patch.

Patch works fine for me:
Kernel command line: BOOT_IMAGE=LinuxNEW ro root=304 idebus=66
ide_setup: idebus=66
ide: Assuming 66MHz system bus speed for PIO modes


Thanks a lot,

-- 
Auzanneau Grégory
GPG 0x99137BEE

Re: idebus setup problem (2.6.7-rc1)

[Bartlomiej Zolnierkiewicz]

On Friday 28 of May 2004 11:11, Auzanneau Gregory wrote:
> Zhu, Yi a écrit :
> > My bad. I missed ide_setup(), please try below patch.

Thanks.

> Patch works fine for me:
> Kernel command line: BOOT_IMAGE=LinuxNEW ro root=304 idebus=66
> ide_setup: idebus=66
> ide: Assuming 66MHz system bus speed for PIO modes

OK, now please explain why do you use 'idebus=66'. :-)

> Thanks a lot,

Re: idebus setup problem (2.6.7-rc1)

[Auzanneau Gregory]

Bartlomiej Zolnierkiewicz a écrit :
> OK, now please explain why do you use 'idebus=66'. :-)

Because my SIS 5513 southbridge seems to run by default to 33Mhz.

-- 
Auzanneau Grégory
GPG 0x99137BEE

Re: idebus setup problem (2.6.7-rc1)

[Bartlomiej Zolnierkiewicz]

On Friday 28 of May 2004 14:52, Auzanneau Gregory wrote:
> Bartlomiej Zolnierkiewicz a écrit :
> > OK, now please explain why do you use 'idebus=66'. :-)
>
> Because my SIS 5513 southbridge seems to run by default to 33Mhz.

and you run PCI bus at 66MHz?

SiS IDE driver completely ignores 'idebus' parameter.

RE: idebus setup problem (2.6.7-rc1)

[Rusty Russell]

On Fri, 2004-05-28 at 02:38, Zhu, Yi wrote:
> linux-kernel-owner@vger.kernel.org wrote:
> > Bartlomiej Zolnierkiewicz wrote:
> >> 
> >> It breaks all "idex=" and "hdx=" options.
> >> Please take a look at how ide_setup().
> > 
> > Yes, thanks for pointing out. Maybe we need some wildcard
> > support. If module_param() can do this, that's great.
> 
> Does below change acceptable to make module_param support 
> wildcard '?' ?

Dislike this idea.  If you have hundreds of parameters, maybe it's
supposed to be a PITA?

Rusty,
-- 
Anyone who quotes me in their signature is an idiot -- Rusty Russell

RE: idebus setup problem (2.6.7-rc1)

[Zhu, Yi]

Rusty Russell wrote:
> 
> Dislike this idea.  If you have hundreds of parameters, maybe it's
> supposed to be a PITA? 

What's your idea to make module_param support alterable param
names like ide3=xxx ?

Thanks,
-yi

Re: idebus setup problem (2.6.7-rc1)

[Herbert Poetzl]

On Thu, Jun 03, 2004 at 10:05:08PM +0800, Zhu, Yi wrote:
> Rusty Russell wrote:
> > 
> > Dislike this idea.  If you have hundreds of parameters, maybe it's
> > supposed to be a PITA? 
> 
> What's your idea to make module_param support alterable param
> names like ide3=xxx ?

hmm, what about making all those something like:

	ide=3:foo,bar;4:wossname

where ':' and ';' are arbitrarily chosen atm ...
or, if it should be handled at the 'argument' level
maybe the 'notion' of arrays would help, something
like
	
	ide[3]=foo,bar

where ide is defined as array of strings, size 16

best,
Herbert

> Thanks,
> -yi
> 
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/

Re: idebus setup problem (2.6.7-rc1)

[Bartlomiej Zolnierkiewicz]

On Thursday 03 of June 2004 23:31, Herbert Poetzl wrote:
> On Thu, Jun 03, 2004 at 10:05:08PM +0800, Zhu, Yi wrote:
> > Rusty Russell wrote:
> > > Dislike this idea.  If you have hundreds of parameters, maybe it's
> > > supposed to be a PITA?
> >
> > What's your idea to make module_param support alterable param
> > names like ide3=xxx ?
>
> hmm, what about making all those something like:
>
> 	ide=3:foo,bar;4:wossname

We are in stable kernel and in 2.7 'idex=' and 'hdx=' will die.

> where ':' and ';' are arbitrarily chosen atm ...
> or, if it should be handled at the 'argument' level
> maybe the 'notion' of arrays would help, something
> like
>
> 	ide[3]=foo,bar
>
> where ide is defined as array of strings, size 16

Sounds awful.

Cheers,
Bartlomiej

Re: idebus setup problem (2.6.7-rc1)

[Rusty Russell]

On Fri, 2004-06-04 at 07:44, Bartlomiej Zolnierkiewicz wrote:
> On Thursday 03 of June 2004 23:31, Herbert Poetzl wrote:
> > On Thu, Jun 03, 2004 at 10:05:08PM +0800, Zhu, Yi wrote:
> > > Rusty Russell wrote:
> > > > Dislike this idea.  If you have hundreds of parameters, maybe it's
> > > > supposed to be a PITA?
> > >
> > > What's your idea to make module_param support alterable param
> > > names like ide3=xxx ?
> >
> > hmm, what about making all those something like:
> >
> > 	ide=3:foo,bar;4:wossname
> 
> We are in stable kernel and in 2.7 'idex=' and 'hdx=' will die.

Yes, and if you want to clean this up for 2.6, I'd recommend simply
putting twenty module_param_call() lines.

It's ugly, but that's because it's doing ugly things, IMHO, and I don't
think Bart would disagree?

Rusty.
-- 
Anyone who quotes me in their signature is an idiot -- Rusty Russell

Re: idebus setup problem (2.6.7-rc1)

[Bartlomiej Zolnierkiewicz]

On Friday 04 of June 2004 06:32, Rusty Russell wrote:
> On Fri, 2004-06-04 at 07:44, Bartlomiej Zolnierkiewicz wrote:
> > On Thursday 03 of June 2004 23:31, Herbert Poetzl wrote:
> > > On Thu, Jun 03, 2004 at 10:05:08PM +0800, Zhu, Yi wrote:
> > > > Rusty Russell wrote:
> > > > > Dislike this idea.  If you have hundreds of parameters, maybe it's
> > > > > supposed to be a PITA?
> > > >
> > > > What's your idea to make module_param support alterable param
> > > > names like ide3=xxx ?
> > >
> > > hmm, what about making all those something like:
> > >
> > > 	ide=3:foo,bar;4:wossname
> >
> > We are in stable kernel and in 2.7 'idex=' and 'hdx=' will die.
>
> Yes, and if you want to clean this up for 2.6, I'd recommend simply
> putting twenty module_param_call() lines.

10 for "idex="
20 for "hdx="
 1 for "idebus="
 1 for "ide="

I tried it once and result was uglier than the current behavior
(part of the problem is that MAX_HWIFS is arch/config dependent).

Why can't we apply this minimal fix from Yi for now?

--- linux-2.6.7-rc1-mm1.orig/init/main.c        2004-05-28 12:39:15.549314064 +0800
+++ linux-2.6.7-rc1-mm1/init/main.c     2004-05-28 12:40:29.399087192 +0800
@@ -162,7 +162,7 @@ static int __init obsolete_checksetup(ch
        p = &__setup_start;
        do {
                int n = strlen(p->str);
-               if (len <= n && !strncmp(line, p->str, n)) {
+               if (n == 0 || (len <= n && !strncmp(line, p->str, n))) {
                        /* Already done in parse_early_param? */
                        if (p->early)
                                return 1;

> It's ugly, but that's because it's doing ugly things, IMHO, and I don't
> think Bart would disagree?
>
> Rusty.

RE: idebus setup problem (2.6.7-rc1)

[Zhu, Yi]

Bartlomiej Zolnierkiewicz wrote:
> 
> Why can't we apply this minimal fix from Yi for now?

Thanks, this is in 2.6.7-rc2-mm2.

> 
> --- linux-2.6.7-rc1-mm1.orig/init/main.c        2004-05-28
> 12:39:15.549314064 +0800 +++ linux-2.6.7-rc1-mm1/init/main.c    
> 2004-05-28 12:40:29.399087192 +0800
> @@ -162,7 +162,7 @@ static int __init obsolete_checksetup(ch        
>         p = &__setup_start; do {
>                 int n = strlen(p->str);
> -               if (len <= n && !strncmp(line, p->str, n)) {
> +               if (n == 0 || (len <= n && !strncmp(line, p->str,
>                         n))) { /* Already done in parse_early_param?
>                         */ if (p->early)
>                                 return 1;
> 

Thanks,
-yi

RE: idebus setup problem (2.6.7-rc1)

[Rusty Russell]

On Sun, 2004-06-06 at 23:11, Zhu, Yi wrote:
> Bartlomiej Zolnierkiewicz wrote:
> > 
> > Why can't we apply this minimal fix from Yi for now?
> 
> Thanks, this is in 2.6.7-rc2-mm2.

OK, I've revisited this problem, with my thinking cap ON this time. 
Sorry for the delay.

Andrew, please revert kernel-parameter-parsing-fix.patch and 
kernel-parameter-parsing-fix-fix.patch in favor of this one-liner.

Yi, does this fix your ACPI problem?

Rusty.

Name: Handle __early_param and __setup Collision
Status: Trivial
Depends: EarlyParam/early_param.patch.gz
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

Yi Zhu (yi.zhu@intel.com) points out the following problem:

In arch/i386/kernel/setup.c:
	__early_param("acpi", early_acpi);

In drivers/acpi/osl.c:
	__setup("acpi_os_name=", acpi_os_name_setup);

The problem command line looks like:

	"acpi=force acpi_os_name=my_override_name"

For simplicity, we overload the __setup section to contain both
__early_param and __setup, so we can check that all options on the
command line are taken by at least one of them.  However,
__early_param have different semantics the __setup: in particular,
__early_param("acpi"), must not match anything but "acpi" and "acpi=",
which mirrors module_param(), whereas __setup("acpi") would match
anything which starts with "acpi".

Fix the obsolete_checksetup code to take this difference into account
correctly.
 
diff -urpN --exclude TAGS -X /home/rusty/devel/kernel/kernel-patches/current-dontdiff --minimal .22424-linux-2.6.7-rc2-bk7/init/main.c .22424-linux-2.6.7-rc2-bk7.updated/init/main.c
--- .22424-linux-2.6.7-rc2-bk7/init/main.c	2004-06-07 09:51:11.000000000 +1000
+++ .22424-linux-2.6.7-rc2-bk7.updated/init/main.c	2004-06-07 09:53:06.000000000 +1000
@@ -159,8 +159,9 @@ static int __init obsolete_checksetup(ch
 	do {
 		int n = strlen(p->str);
 		if (!strncmp(line, p->str, n)) {
-			/* Already done in parse_early_param? */
-			if (p->early)
+			/* Already done in parse_early_param?  (Needs
+			 * exact match on param part) */
+			if (p->early && (line[n] == '\0' || line[n] == '='))
 				return 1;
 			if (!p->setup_func) {
 				printk(KERN_WARNING "Parameter %s is obsolete, ignored\n", p->str);

-- 
Anyone who quotes me in their signature is an idiot -- Rusty Russell

RE: idebus setup problem (2.6.7-rc1)

[Zhu, Yi]

Rusty Russell wrote:
> 
> OK, I've revisited this problem, with my thinking cap ON this time.
> Sorry for the delay. 
> 
> Andrew, please revert kernel-parameter-parsing-fix.patch and
> kernel-parameter-parsing-fix-fix.patch in favor of this one-liner.
> 
> Yi, does this fix your ACPI problem?
> 
> Rusty.
> 
> Name: Handle __early_param and __setup Collision
> Status: Trivial
> Depends: EarlyParam/early_param.patch.gz
> Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
> 
> Yi Zhu (yi.zhu@intel.com) points out the following problem:
> 
> In arch/i386/kernel/setup.c:
> 	__early_param("acpi", early_acpi);
> 
> In drivers/acpi/osl.c:
> 	__setup("acpi_os_name=", acpi_os_name_setup);
> 
> The problem command line looks like:
> 
> 	"acpi=force acpi_os_name=my_override_name"
> 
> For simplicity, we overload the __setup section to contain
> both __early_param and __setup, so we can check that all
> options on the command line are taken by at least one of
> them.  However, __early_param have different semantics the
> __setup: in particular, __early_param("acpi"), must not match
> anything but "acpi" and "acpi=", which mirrors
> module_param(), whereas __setup("acpi") would match anything
> which starts with "acpi".

Really? I think currently only ide_setup is an exception for __setup(),
which will match all params given in command line.

> 
> Fix the obsolete_checksetup code to take this difference into account
> correctly. 
> 
> diff -urpN --exclude TAGS -X
> /home/rusty/devel/kernel/kernel-patches/current-dontdiff
> --minimal .22424-linux-2.6.7-rc2-bk7/init/main.c
> .22424-linux-2.6.7-rc2-bk7.updated/init/main.c
> --- .22424-linux-2.6.7-rc2-bk7/init/main.c	2004-06-07
> 09:51:11.000000000 +1000 +++
> .22424-linux-2.6.7-rc2-bk7.updated/init/main.c 2004-06-07
> 09:53:06.000000000 +1000 @@ -159,8 +159,9 @@ static int __init
>  		obsolete_checksetup(ch  	do { int n =
strlen(p->str);
>  		if (!strncmp(line, p->str, n)) {
> -			/* Already done in parse_early_param? */
> -			if (p->early)
> +			/* Already done in parse_early_param?  (Needs
> +			 * exact match on param part) */
> +			if (p->early && (line[n] == '\0' ||
> line[n] == '='))
>  				return 1;
>  			if (!p->setup_func) {
>  				printk(KERN_WARNING "Parameter
> %s is obsolete, ignored\n", p->str);

This doesn't work. The p->setup_func for "acpi" will still be called on
behalf of "acpi_os_name".

Below change should work.

Thanks,
-yi

--- linux-2.6.7-rc2-mm2.orig/init/main.c	2004-06-07
16:01:25.000000000 +0800
+++ linux-2.6.7-rc2-mm2/init/main.c	2004-06-07 18:50:22.256204536
+0800
@@ -154,18 +154,22 @@ static int __init obsolete_checksetup(ch
 {
 	struct obs_kernel_param *p;
 	extern struct obs_kernel_param __setup_start, __setup_end;
-	char *ptr;
-	int len = strlen(line);
 
-	if ((ptr = strchr(line, '=')))
-		len = ptr - line;
 	p = &__setup_start;
 	do {
 		int n = strlen(p->str);
-		if (n == 0 || (len <= n && !strncmp(line, p->str, n))) {
-			/* Already done in parse_early_param? */
-			if (p->early)
-				return 1;
+		if (!strncmp(line, p->str, n)) {
+			if (p->early) {
+				/* Already done in parse_early_param?
(Needs
+				 * exact match on param part) */
+				if (p->early && (line[n] == '\0' ||
+				    line[n] == '='))
+					return 1;
+				else {
+					p++;
+					continue;
+				}
+			}
 			if (!p->setup_func) {
 				printk(KERN_WARNING "Parameter %s is
obsolete,"
 						" ignored\n", p->str);

RE: idebus setup problem (2.6.7-rc1)

[Rusty Russell]

On Mon, 2004-06-07 at 21:04, Zhu, Yi wrote:
> > For simplicity, we overload the __setup section to contain
> > both __early_param and __setup, so we can check that all
> > options on the command line are taken by at least one of
> > them.  However, __early_param have different semantics the
> > __setup: in particular, __early_param("acpi"), must not match
> > anything but "acpi" and "acpi=", which mirrors
> > module_param(), whereas __setup("acpi") would match anything
> > which starts with "acpi".
> 
> Really? I think currently only ide_setup is an exception for __setup(),
> which will match all params given in command line.

No, there is at least one other place which relies on this feature last
I searched.  They're all supposed to be stem matches, so __setup("foo")
matches "foobar=100" for example.

> > -			/* Already done in parse_early_param? */
> > -			if (p->early)
> > +			/* Already done in parse_early_param?  (Needs
> > +			 * exact match on param part) */
> > +			if (p->early && (line[n] == '\0' ||
> > line[n] == '='))
> >  				return 1;
> >  			if (!p->setup_func) {
> >  				printk(KERN_WARNING "Parameter
> > %s is obsolete, ignored\n", p->str);
> 
> This doesn't work. The p->setup_func for "acpi" will still be called on
> behalf of "acpi_os_name".
> 
> Below change should work.

Good point, but I think this is clearer:

Name: Handle __early_param and __setup Collision
Status: Booted on 2.6.7-rc2-bk7
Depends: EarlyParam/early_param.patch.gz
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

Yi Zhu (yi.zhu@intel.com) points out the following problem:

In arch/i386/kernel/setup.c:
	__early_param("acpi", early_acpi);

In drivers/acpi/osl.c:
	__setup("acpi_os_name=", acpi_os_name_setup);

The problem command line looks like:

	"acpi=force acpi_os_name=my_override_name"

For simplicity, we overload the __setup section to contain both
__early_param and __setup, so we can check that all options on the
command line are taken by at least one of them.  However,
__early_param have different semantics the __setup: in particular,
__early_param("acpi"), must not match anything but "acpi" and "acpi=",
which mirrors module_param(), whereas __setup("acpi") would match
anything which starts with "acpi".

Fix the obsolete_checksetup code to take this difference into account
correctly.
 
diff -urpN --exclude TAGS -X /home/rusty/devel/kernel/kernel-patches/current-dontdiff --minimal .9590-linux-2.6.7-rc2-bk7/init/main.c .9590-linux-2.6.7-rc2-bk7.updated/init/main.c
--- .9590-linux-2.6.7-rc2-bk7/init/main.c	2004-06-08 08:44:23.000000000 +1000
+++ .9590-linux-2.6.7-rc2-bk7.updated/init/main.c	2004-06-08 08:48:15.000000000 +1000
@@ -159,10 +159,12 @@ static int __init obsolete_checksetup(ch
 	do {
 		int n = strlen(p->str);
 		if (!strncmp(line, p->str, n)) {
-			/* Already done in parse_early_param? */
-			if (p->early)
-				return 1;
-			if (!p->setup_func) {
+			if (p->early) {
+				/* Already done in parse_early_param?  (Needs
+				 * exact match on param part) */
+				if (line[n] == '\0' || line[n] == '=')
+					return 1;
+			} else if (!p->setup_func) {
 				printk(KERN_WARNING "Parameter %s is obsolete, ignored\n", p->str);
 				return 1;
 			} else if (p->setup_func(line + n))

-- 
Anyone who quotes me in their signature is an idiot -- Rusty Russell