From: Denys Vlasenko <vda.linux@googlemail.com>
To: Andrew Morton <akpm@linux-foundation.org>
Cc: Matthew Wilcox <matthew@wil.cx>,
Linus Torvalds <torvalds@linux-foundation.org>,
Peter Anvin <hpa@zytor.com>,
"David S. Miller" <davem@davemloft.net>,
linux-ia64@vger.kernel.org, linuxppc-dev@ozlabs.org,
linux-kernel@vger.kernel.org
Subject: Re: the printk problem
Date: Sat, 05 Jul 2008 10:20:25 +0000 [thread overview]
Message-ID: <200807051220.25418.vda.linux@googlemail.com> (raw)
In-Reply-To: <20080704150100.1f7b8a65.akpm@linux-foundation.org>
On Saturday 05 July 2008 00:01, Andrew Morton wrote:
> > > We also jump through hoops to print things like sector_t and
> > > resource_size_t. They always need to be cast to `unsiged long long',
> > > which generates additional stack space and text in some setups.
> >
> > The thing is that GCC checks types. So it's fine to add "print this
> > pointer specially", but you can't in general add new printf arguments
> > without also hacking GCC. Unless you use -Wno-format, and require
> > sparse to check special kernel types.
>
> It would be excellent if gcc had an extension system so that you could
> add new printf control chars and maybe even tell gcc how to check them.
> But of course, if that were to happen, we couldn't use it for 4-5 years.
>
> What I had initially proposed was to abuse %S, which takes a wchar_t*.
> gcc accepts `unsigned long *' for %S.
>
> Then, we put the kernel-specific control char after the S, so we can
> print an inode (rofl) with
>
> struct inode *inode;
>
> printk("here is an inode: %Si\n", (unsigned long *)inode);
>
> Downsides are:
>
> - there's a cast, so you could accidentally do
>
> printk("here is an inode: %Si\n", (unsigned long *)dentry);
>
> - there's a cast, and they're ugly
>
> - gcc cannot of course check that the arg matches the control string
>
> Unfortunately (and this seems weird), gcc printf checking will not
> accept a void* for %S: it _has_ to be wchar_t*, and the checker won't
> permit void* substitution for that.
Repeating myself here...
We can add an alternative alias to printk:
asmlinkage int printk(const char * fmt, ...)
__attribute__ ((format (printf, 1, 2))) __cold;
+asmlinkage int custom_printk(const char * fmt, ...) __cold asm ("printk");
custom_printk() is actually just printk(), that is,
we won't need additional function, we need to teach
*printk* about MAC addresses, NIPQUADs etc;
and then use printk() if you use only standard %fmt (and have it
checked by gcc), or use custom_printk() if you have non-standard
%fmt in the format string.
The only downside that in second case, you lose gcc checking.
No big deal.
--
vda
WARNING: multiple messages have this Message-ID (diff)
From: Denys Vlasenko <vda.linux@googlemail.com>
To: Andrew Morton <akpm@linux-foundation.org>
Cc: linux-ia64@vger.kernel.org, Matthew Wilcox <matthew@wil.cx>,
linux-kernel@vger.kernel.org, linuxppc-dev@ozlabs.org,
Peter Anvin <hpa@zytor.com>,
Linus Torvalds <torvalds@linux-foundation.org>,
"David S. Miller" <davem@davemloft.net>
Subject: Re: the printk problem
Date: Sat, 5 Jul 2008 12:20:25 +0200 [thread overview]
Message-ID: <200807051220.25418.vda.linux@googlemail.com> (raw)
In-Reply-To: <20080704150100.1f7b8a65.akpm@linux-foundation.org>
On Saturday 05 July 2008 00:01, Andrew Morton wrote:
> > > We also jump through hoops to print things like sector_t and
> > > resource_size_t. They always need to be cast to `unsiged long long',
> > > which generates additional stack space and text in some setups.
> >
> > The thing is that GCC checks types. So it's fine to add "print this
> > pointer specially", but you can't in general add new printf arguments
> > without also hacking GCC. Unless you use -Wno-format, and require
> > sparse to check special kernel types.
>
> It would be excellent if gcc had an extension system so that you could
> add new printf control chars and maybe even tell gcc how to check them.
> But of course, if that were to happen, we couldn't use it for 4-5 years.
>
> What I had initially proposed was to abuse %S, which takes a wchar_t*.
> gcc accepts `unsigned long *' for %S.
>
> Then, we put the kernel-specific control char after the S, so we can
> print an inode (rofl) with
>
> struct inode *inode;
>
> printk("here is an inode: %Si\n", (unsigned long *)inode);
>
> Downsides are:
>
> - there's a cast, so you could accidentally do
>
> printk("here is an inode: %Si\n", (unsigned long *)dentry);
>
> - there's a cast, and they're ugly
>
> - gcc cannot of course check that the arg matches the control string
>
> Unfortunately (and this seems weird), gcc printf checking will not
> accept a void* for %S: it _has_ to be wchar_t*, and the checker won't
> permit void* substitution for that.
Repeating myself here...
We can add an alternative alias to printk:
asmlinkage int printk(const char * fmt, ...)
__attribute__ ((format (printf, 1, 2))) __cold;
+asmlinkage int custom_printk(const char * fmt, ...) __cold asm ("printk");
custom_printk() is actually just printk(), that is,
we won't need additional function, we need to teach
*printk* about MAC addresses, NIPQUADs etc;
and then use printk() if you use only standard %fmt (and have it
checked by gcc), or use custom_printk() if you have non-standard
%fmt in the format string.
The only downside that in second case, you lose gcc checking.
No big deal.
--
vda
WARNING: multiple messages have this Message-ID (diff)
From: Denys Vlasenko <vda.linux@googlemail.com>
To: Andrew Morton <akpm@linux-foundation.org>
Cc: Matthew Wilcox <matthew@wil.cx>,
Linus Torvalds <torvalds@linux-foundation.org>,
Peter Anvin <hpa@zytor.com>,
"David S. Miller" <davem@davemloft.net>,
linux-ia64@vger.kernel.org, linuxppc-dev@ozlabs.org,
linux-kernel@vger.kernel.org
Subject: Re: the printk problem
Date: Sat, 5 Jul 2008 12:20:25 +0200 [thread overview]
Message-ID: <200807051220.25418.vda.linux@googlemail.com> (raw)
In-Reply-To: <20080704150100.1f7b8a65.akpm@linux-foundation.org>
On Saturday 05 July 2008 00:01, Andrew Morton wrote:
> > > We also jump through hoops to print things like sector_t and
> > > resource_size_t. They always need to be cast to `unsiged long long',
> > > which generates additional stack space and text in some setups.
> >
> > The thing is that GCC checks types. So it's fine to add "print this
> > pointer specially", but you can't in general add new printf arguments
> > without also hacking GCC. Unless you use -Wno-format, and require
> > sparse to check special kernel types.
>
> It would be excellent if gcc had an extension system so that you could
> add new printf control chars and maybe even tell gcc how to check them.
> But of course, if that were to happen, we couldn't use it for 4-5 years.
>
> What I had initially proposed was to abuse %S, which takes a wchar_t*.
> gcc accepts `unsigned long *' for %S.
>
> Then, we put the kernel-specific control char after the S, so we can
> print an inode (rofl) with
>
> struct inode *inode;
>
> printk("here is an inode: %Si\n", (unsigned long *)inode);
>
> Downsides are:
>
> - there's a cast, so you could accidentally do
>
> printk("here is an inode: %Si\n", (unsigned long *)dentry);
>
> - there's a cast, and they're ugly
>
> - gcc cannot of course check that the arg matches the control string
>
> Unfortunately (and this seems weird), gcc printf checking will not
> accept a void* for %S: it _has_ to be wchar_t*, and the checker won't
> permit void* substitution for that.
Repeating myself here...
We can add an alternative alias to printk:
asmlinkage int printk(const char * fmt, ...)
__attribute__ ((format (printf, 1, 2))) __cold;
+asmlinkage int custom_printk(const char * fmt, ...) __cold asm ("printk");
custom_printk() is actually just printk(), that is,
we won't need additional function, we need to teach
*printk* about MAC addresses, NIPQUADs etc;
and then use printk() if you use only standard %fmt (and have it
checked by gcc), or use custom_printk() if you have non-standard
%fmt in the format string.
The only downside that in second case, you lose gcc checking.
No big deal.
--
vda
next prev parent reply other threads:[~2008-07-05 10:20 UTC|newest]
Thread overview: 111+ messages / expand[flat|nested] mbox.gz Atom feed top
2008-06-25 13:11 [PATCH] handle failure of irqchip->set_type in setup_irq Uwe Kleine-König
2008-07-02 9:17 ` Uwe Kleine-König
2008-07-02 9:49 ` Andrew Morton
2008-07-02 10:09 ` Uwe Kleine-König
2008-07-04 10:46 ` [PATCH v2] " Uwe Kleine-König
2008-07-04 17:17 ` Andrew Morton
2008-07-04 18:43 ` Uwe Kleine-König
2008-07-04 19:08 ` Andrew Morton
2008-07-09 13:13 ` Uwe Kleine-König
2008-07-09 21:52 ` Andrew Morton
2008-07-10 8:23 ` Uwe Kleine-König
2008-07-10 8:28 ` Andrew Morton
[not found] ` <20080704111540.ddffd241.akpm@linux-foundation.org>
[not found] ` <alpine.LFD.1.10.0807041147450.2815@woody.linux-foundation.org>
2008-07-04 20:02 ` the printk problem Linus Torvalds
2008-07-04 20:02 ` Linus Torvalds
2008-07-04 20:27 ` Andrew Morton
2008-07-04 20:27 ` Andrew Morton
2008-07-04 20:41 ` Linus Torvalds
2008-07-04 20:41 ` Linus Torvalds
2008-07-04 20:42 ` Matthew Wilcox
2008-07-04 20:42 ` Matthew Wilcox
2008-07-04 22:01 ` Andrew Morton
2008-07-04 22:01 ` Andrew Morton
2008-07-04 22:01 ` Andrew Morton
2008-07-05 2:03 ` Matthew Wilcox
2008-07-05 2:03 ` Matthew Wilcox
2008-07-05 2:03 ` Matthew Wilcox
2008-07-22 10:05 ` [PATCH] Make u64 long long on all architectures (was: the printk problem) Andrew Morton
2008-07-22 10:05 ` Andrew Morton
2008-07-22 10:05 ` [PATCH] Make u64 long long on all architectures (was: the Andrew Morton
2008-07-22 10:36 ` [PATCH] Make u64 long long on all architectures (was: the printk problem) Michael Ellerman
2008-07-22 10:36 ` Michael Ellerman
2008-07-22 10:36 ` [PATCH] Make u64 long long on all architectures (was: the Michael Ellerman
2008-07-22 10:53 ` [PATCH] Make u64 long long on all architectures (was: the printk problem) Andrew Morton
2008-07-22 10:53 ` Andrew Morton
2008-07-22 10:53 ` [PATCH] Make u64 long long on all architectures (was: the Andrew Morton
2008-07-22 11:36 ` [PATCH] Make u64 long long on all architectures (was: the printk problem) Benjamin Herrenschmidt
2008-07-22 11:36 ` Benjamin Herrenschmidt
2008-07-22 11:36 ` [PATCH] Make u64 long long on all architectures (was: the Benjamin Herrenschmidt
2008-07-22 11:35 ` [PATCH] Make u64 long long on all architectures (was: the printk problem) Benjamin Herrenschmidt
2008-07-22 11:35 ` Benjamin Herrenschmidt
2008-07-22 11:35 ` [PATCH] Make u64 long long on all architectures (was: the Benjamin Herrenschmidt
2008-07-05 10:20 ` Denys Vlasenko [this message]
2008-07-05 10:20 ` the printk problem Denys Vlasenko
2008-07-05 10:20 ` Denys Vlasenko
2008-07-05 11:33 ` Jan Engelhardt
2008-07-05 11:33 ` Jan Engelhardt
2008-07-05 11:33 ` Jan Engelhardt
2008-07-04 22:58 ` Benjamin Herrenschmidt
2008-07-04 22:58 ` Benjamin Herrenschmidt
2008-07-04 20:36 ` Matthew Wilcox
2008-07-04 20:36 ` Matthew Wilcox
2008-07-08 1:44 ` Kyle McMartin
2008-07-08 1:44 ` Kyle McMartin
2008-07-04 23:00 ` Benjamin Herrenschmidt
2008-07-04 23:00 ` Benjamin Herrenschmidt
2008-07-04 23:25 ` Linus Torvalds
2008-07-04 23:25 ` Linus Torvalds
2008-07-05 22:32 ` Linus Torvalds
2008-07-05 22:32 ` Linus Torvalds
2008-07-05 22:57 ` Arjan van de Ven
2008-07-05 22:57 ` Arjan van de Ven
2008-07-06 5:27 ` Ingo Molnar
2008-07-06 5:27 ` Ingo Molnar
2008-07-06 5:37 ` Linus Torvalds
2008-07-06 5:37 ` Linus Torvalds
2008-07-06 5:37 ` Linus Torvalds
2008-07-06 5:53 ` Ingo Molnar
2008-07-06 5:53 ` Ingo Molnar
2008-07-06 5:53 ` Ingo Molnar
2008-07-06 6:13 ` Ingo Molnar
2008-07-06 6:13 ` Ingo Molnar
2008-07-06 6:13 ` Ingo Molnar
2008-07-07 1:14 ` Benjamin Herrenschmidt
2008-07-07 1:14 ` Benjamin Herrenschmidt
2008-07-07 3:26 ` Stephen Rothwell
2008-07-07 3:28 ` Michael Ellerman
2008-07-07 4:59 ` Stephen Rothwell
2008-07-07 3:43 ` Benjamin Herrenschmidt
2008-07-05 12:52 ` Vegard Nossum
2008-07-05 12:52 ` Vegard Nossum
2008-07-05 12:52 ` Vegard Nossum
2008-07-05 13:24 ` Jan Engelhardt
2008-07-05 13:24 ` Jan Engelhardt
2008-07-05 13:24 ` Jan Engelhardt
2008-07-05 13:50 ` Vegard Nossum
2008-07-05 13:50 ` Vegard Nossum
2008-07-05 13:50 ` Vegard Nossum
2008-07-05 14:07 ` Jan Engelhardt
2008-07-05 14:07 ` Jan Engelhardt
2008-07-05 14:07 ` Jan Engelhardt
2008-07-05 17:56 ` Linus Torvalds
2008-07-05 17:56 ` Linus Torvalds
2008-07-05 17:56 ` Linus Torvalds
2008-07-05 18:40 ` Jan Engelhardt
2008-07-05 18:40 ` Jan Engelhardt
2008-07-05 18:40 ` Jan Engelhardt
2008-07-05 18:44 ` Linus Torvalds
2008-07-05 18:44 ` Linus Torvalds
2008-07-05 18:44 ` Linus Torvalds
2008-07-05 18:41 ` Vegard Nossum
2008-07-05 18:41 ` Vegard Nossum
2008-07-05 18:41 ` Vegard Nossum
2008-07-05 18:52 ` Matthew Wilcox
2008-07-05 18:52 ` Matthew Wilcox
2008-07-05 18:52 ` Matthew Wilcox
2008-07-06 0:02 ` Pekka Enberg
2008-07-06 0:02 ` Pekka Enberg
2008-07-06 0:02 ` Pekka Enberg
2008-07-06 5:17 ` Randy Dunlap
2008-07-06 5:17 ` Randy Dunlap
2008-07-06 5:17 ` Randy Dunlap
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=200807051220.25418.vda.linux@googlemail.com \
--to=vda.linux@googlemail.com \
--cc=akpm@linux-foundation.org \
--cc=davem@davemloft.net \
--cc=hpa@zytor.com \
--cc=linux-ia64@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linuxppc-dev@ozlabs.org \
--cc=matthew@wil.cx \
--cc=torvalds@linux-foundation.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.