From: Jim Cromie <jim.cromie@gmail.com>
To: Joe Perches <joe@perches.com>
Cc: jbaron@redhat.com, bart.vanassche@gmail.com, greg@kroah.com,
linux-kernel@vger.kernel.org
Subject: Re: [PATCH 25/26] dynamic_debug: add pr_fmt_dbg() for dynamic_pr_debug
Date: Wed, 28 Sep 2011 11:27:09 -0600 [thread overview]
Message-ID: <CAJfuBxxoJ=KvzS-XgD6h7oXZ==xOC72_1YiR6shkY=5wC7CNmw@mail.gmail.com> (raw)
In-Reply-To: <1317185489.19340.22.camel@Joe-Laptop>
On Tue, Sep 27, 2011 at 10:51 PM, Joe Perches <joe@perches.com> wrote:
> On Tue, 2011-09-27 at 20:54 -0600, Jim Cromie wrote:
>> Cleanest way is to drop the original synonym, and just use
>> warn (its shorter), but that creates some churn (havent grepped to see
>> how much).
>> I picked what looked like least effort & fewest corner-cases.
>> ICBW..
>
> #ifndef pr_fmt_warning
> #define pr_fmt_warning pr_fmt_warn
> #endif
>
you've made it conditional, where line 203 is not. why ?
201 #define pr_warning(fmt, ...) \
202 printk(KERN_WARNING pr_fmt_warn(fmt), ##__VA_ARGS__)
203 #define pr_warn pr_warning
204 #define pr_notice(fmt, ...) \
Also, though minor, 203 adds pr_warn as the shortcut,
youve suggested the opposite. I presume this is an oversight.
>> > What did you think of avoiding all of this and
>> > having __dynamic_pr_debug move the fmt pointer over
>> > any initial KBUILD_MODULE ": "
>> >
>> > int __dynamic_pr_debug(struct _ddebug *descriptor, const char *fmt, ...)
>> > {
>> > [...]
>> > size_t modsize = strlen(descriptor->modname);
>> > if (0 == strncmp(fmt, descriptor->modname, modsize) &&
>> > 0 == strncmp(fmt + modsize, ": ", 2))
>> > fmt += modsize + 2;
>> > vprintk(fmt, ...)
>> > ?
>>
>> I was getting to that... ;-)
>> Im not crazy about it. It feels like too much ..
>> Its a runtime workaround for what I think is a
>> problem in users' (or header's) #defines.
>
> I think exactly the opposite myself.
>
> I think all of the '#define pr_fmt(fmt) KBUILD_MODNAME ": "'
> are effectivly useless and I will eventually delete them.
>
> The printk subsystem should look up the module name and
> prefix them to the printk akin to how the dynamic_debug
> system does. Except the module name should be a singleton.
hmm, maybe Ive missed something in your argument.
For non-dynamic-debug builds, it seems you still want
the MODNAME in the pr_(crit|warn|info|debug) output,
you just dont like the hundreds of defines throughout drivers/*
linux-2.6]$ grep -r pr_fmt drivers/ |wc
427 2556 32040
linux-2.6]$ grep -r pr_fmt drivers/ | grep KBUILD_MODNAME |wc
303 1827 22567
Your runtime check-and-skip-MODNAME code above
fixes dynamic-debug so that works around user defines
doing:
#define pr_fmt(fmt) KBUILD_MODNAME ": "
This would let user of a dynamic-debug enabled kernel
add MODNAME selectively, with +m >$CONTROL
But ISTM this does the same:
diff --git a/include/linux/printk.h b/include/linux/printk.h
index e21de7e..cf17986 100644
--- a/include/linux/printk.h
+++ b/include/linux/printk.h
@@ -148,8 +148,17 @@ static inline void setup_log_buf(int early)
extern void dump_stack(void) __cold;
+/* Set pr_fmt default to most common, useful case */
#ifndef pr_fmt
-#define pr_fmt(fmt) fmt
+#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
+#endif
+/* except for dynamic-debug, where user can enable it, and we dont
+ want to print it 2x
+ */
+#if defined(CONFIG_DYNAMIC_DEBUG)
+#ifndef pr_fmt_debug
+#define pr_fmt_debug(fmt) fmt
+#endif
#endif
#ifndef pr_fmt_emerg
This sets default to work for the 303 common cases,
allowing the removal of their defines:
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
It also allows existing non-standard uses to work as is:
drivers/sh/intc/dynamic.c:#define pr_fmt(fmt) "intc: " fmt
Your suggested pr_fmt_warning, etc allow severity specific
customizations to override, w/o changing all the others..
And in dynamic-debug kernels, the default pr_fmt_debug
is reset so MODNAME isnt printed, except by the user:
~]# echo +m > $CONTROL
IOW, it seems to take care of everything, w/o runtime workarounds.
With agreement on this, I can fold above patch into 25/26
(which Ive updated here to also fix pr_*_once, pr_*_ratelimited macros),
and we can move against those 303 irritants at our leisure (there is
no flag day)
thanks
Jim
next prev parent reply other threads:[~2011-09-28 17:27 UTC|newest]
Thread overview: 82+ messages / expand[flat|nested] mbox.gz Atom feed top
2011-09-21 21:54 jim.cromie
2011-09-21 21:54 ` [PATCH 01/26] dynamic_debug: drop enabled field from struct _ddebug, use flags & _DPRINTK_FLAGS_PRINT jim.cromie
2011-09-21 21:54 ` [PATCH 02/26] dynamic_debug: make dynamic-debug supersede DEBUG ccflag jim.cromie
2011-09-21 21:54 ` [PATCH 03/26] dynamic_debug: replace strcpy with strlcpy, in ddebug_setup_query() jim.cromie
2011-09-21 21:54 ` [PATCH 04/26] dynamic_debug: warn when >1 of each type of match-spec is given jim.cromie
2011-09-21 21:54 ` [PATCH 05/26] dynamic_debug: pr_err() call should not depend upon verbosity jim.cromie
2011-09-21 21:54 ` [PATCH 06/26] dynamic_debug: add trim_prefix() to provide source-root relative paths jim.cromie
2011-09-21 21:54 ` [PATCH 07/26] dynamic_debug: change verbosity at runtime jim.cromie
2011-09-21 21:54 ` [PATCH 08/26] dynamic_debug: define several levels of verbosity jim.cromie
2011-09-21 21:54 ` [PATCH 09/26] dynamic_debug: process multiple commands on a line jim.cromie
2011-09-21 21:54 ` [PATCH 10/26] dynamic_debug: enlarge command/query write buffer jim.cromie
2011-09-21 22:29 ` Joe Perches
2011-09-21 21:55 ` [PATCH 11/26] dynamic_debug: hoist locking in ddebug_change to callers jim.cromie
2011-09-21 21:55 ` [PATCH 12/26] dynamic_debug: dont kill entire facility on error parsing ddebug_query jim.cromie
2011-09-21 21:55 ` [PATCH 13/26] dynamic_debug: factor vpr_info_dq out of ddebug_parse_query jim.cromie
2011-09-21 21:55 ` [PATCH 14/26] dynamic_debug: refactor query_matches_callsite out of ddebug_change jim.cromie
2011-09-21 21:55 ` [PATCH 15/26] dynamic_debug: drop explicit foo != NULL checks jim.cromie
2011-09-21 21:55 ` [PATCH 16/26] dynamic_debug: save 'a' queries to pending-list for later (re)application jim.cromie
2011-09-21 21:55 ` [PATCH 17/26] dynamic_debug: remove pending query when flags zeroed jim.cromie
2011-09-21 21:55 ` [PATCH 18/26] dynamic_debug: describe_flags with '=[pmflta_]*' jim.cromie
2011-09-21 21:55 ` [PATCH 19/26] dynamic_debug: add flags filtering to flags spec jim.cromie
2011-09-21 21:55 ` [PATCH 20/26] dynamic_debug: make ddebug_describe_flags more generic jim.cromie
2011-09-21 21:55 ` [PATCH 21/26] dynamic_debug: add $DBGFS/dynamic_debug/pending file jim.cromie
2011-09-21 21:55 ` [PATCH 22/26] dynamic_debug: early return if _ddebug table is empty jim.cromie
2011-09-21 21:55 ` [PATCH 23/26] dynamic_debug: document pending queries, flags-filter, multiple queries jim.cromie
2011-09-21 22:35 ` Randy Dunlap
2011-09-27 16:33 ` [patch 23/26 revised] jim.cromie
2011-09-27 16:33 ` [PATCH 23/26] dynamic_debug: document pending queries, flags-filter, multiple queries jim.cromie
2011-10-10 18:54 ` Jonathan Corbet
2011-10-10 19:27 ` Jason Baron
2011-10-18 20:41 ` Jim Cromie
2011-10-19 20:52 ` Jason Baron
2011-09-21 21:55 ` [PATCH 24/26] dynamic_debug: reduce lineno field to a saner 18 bits jim.cromie
2011-09-21 21:55 ` [PATCH 25/26] dynamic_debug: add pr_fmt_dbg() for dynamic_pr_debug jim.cromie
2011-09-22 20:57 ` Joe Perches
2011-09-23 10:31 ` Bart Van Assche
2011-09-23 17:42 ` Joe Perches
2011-09-27 20:37 ` Jim Cromie
2011-09-27 23:36 ` Joe Perches
2011-09-28 2:54 ` Jim Cromie
2011-09-28 4:51 ` Joe Perches
2011-09-28 17:27 ` Jim Cromie [this message]
2011-09-29 15:24 ` Joe Perches
2011-10-08 19:07 ` Bart Van Assche
2011-10-08 20:53 ` Joe Perches
2011-09-27 21:40 ` [revised patches 25,26/26] dynamic_debug: add pr_fmt_*() jim.cromie
2011-09-27 21:40 ` [PATCH 25/26] dynamic_debug: add pr_fmt_debug() for dynamic_pr_debug jim.cromie
2011-09-27 21:41 ` [PATCH 26/26] scx200_acb: use pr_(info|warn|err|debug) and dynamic-debug facilities jim.cromie
2011-09-21 21:55 ` jim.cromie
2011-09-26 23:23 ` your mail Greg KH
2011-10-07 20:33 ` [PATCH 00/26] dynamic_debug: add pending queries, etc (~resend) jim.cromie
2011-10-07 20:33 ` [PATCH 01/26] dynamic_debug: drop enabled field from struct _ddebug, use flags & _DPRINTK_FLAGS_PRINT jim.cromie
2011-10-07 20:33 ` [PATCH 02/26] dynamic_debug: make dynamic-debug supersede DEBUG ccflag jim.cromie
2011-10-07 20:33 ` [PATCH 03/26] dynamic_debug: replace strcpy with strlcpy, in ddebug_setup_query() jim.cromie
2011-10-07 20:33 ` [PATCH 04/26] dynamic_debug: warn when >1 of each type of match-spec is given jim.cromie
2011-10-07 20:33 ` [PATCH 05/26] dynamic_debug: pr_err() call should not depend upon verbosity jim.cromie
2011-10-07 21:21 ` Joe Perches
2011-10-07 20:33 ` [PATCH 06/26] dynamic_debug: add trim_prefix() to provide source-root relative paths jim.cromie
2011-10-07 20:33 ` [PATCH 07/26] dynamic_debug: change verbosity at runtime jim.cromie
2011-10-07 20:33 ` [PATCH 08/26] dynamic_debug: define several levels of verbosity jim.cromie
2011-10-07 20:33 ` [PATCH 09/26] dynamic_debug: process multiple commands on a line jim.cromie
2011-10-07 20:33 ` [PATCH 10/26] dynamic_debug: enlarge command/query write buffer jim.cromie
2011-10-07 20:33 ` [PATCH 11/26] dynamic_debug: hoist locking in ddebug_change to callers jim.cromie
2011-10-07 20:33 ` [PATCH 12/26] dynamic_debug: dont kill entire facility on error parsing ddebug_query jim.cromie
2011-10-07 20:33 ` [PATCH 13/26] dynamic_debug: factor vpr_info_dq out of ddebug_parse_query jim.cromie
2011-10-07 20:33 ` [PATCH 14/26] dynamic_debug: refactor query_matches_callsite out of ddebug_change jim.cromie
2011-10-07 20:33 ` [PATCH 15/26] dynamic_debug: drop explicit foo != NULL checks jim.cromie
2011-10-07 20:33 ` [PATCH 16/26] dynamic_debug: save 'a' queries to pending-list for later (re)application jim.cromie
2011-10-07 20:33 ` [PATCH 17/26] dynamic_debug: remove pending query when flags zeroed jim.cromie
2011-10-07 20:33 ` [PATCH 18/26] dynamic_debug: describe_flags with '=[pmflta_]*' jim.cromie
2011-10-07 20:33 ` [PATCH 19/26] dynamic_debug: add flags filtering to flags spec jim.cromie
2011-10-07 20:33 ` [PATCH 20/26] dynamic_debug: make ddebug_describe_flags more generic jim.cromie
2011-10-07 20:33 ` [PATCH 21/26] dynamic_debug: add $DBGFS/dynamic_debug/pending file jim.cromie
2011-10-07 20:33 ` [PATCH 22/26] dynamic_debug: early return if _ddebug table is empty jim.cromie
2011-10-07 20:33 ` [PATCH 23/26] dynamic_debug: document pending queries, flags-filter, multiple queries jim.cromie
2011-10-07 20:33 ` [PATCH 24/26] dynamic_debug: reduce lineno field to a saner 18 bits jim.cromie
2011-10-07 20:33 ` [PATCH 25/26] dynamic_debug: add pr_fmt_*() for each severity jim.cromie
2011-10-07 20:33 ` [PATCH 26/26] printk.h: fix pr_cont_once() to not use pr_fmt() jim.cromie
2011-10-07 20:45 ` Joe Perches
2011-10-18 18:52 ` [PATCH 00/26] dynamic_debug: add pending queries, etc (~resend) Greg KH
2011-10-19 20:57 ` Jason Baron
2011-10-19 22:29 ` Greg KH
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='CAJfuBxxoJ=KvzS-XgD6h7oXZ==xOC72_1YiR6shkY=5wC7CNmw@mail.gmail.com' \
--to=jim.cromie@gmail.com \
--cc=bart.vanassche@gmail.com \
--cc=greg@kroah.com \
--cc=jbaron@redhat.com \
--cc=joe@perches.com \
--cc=linux-kernel@vger.kernel.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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).