Thread (27 messages) 27 messages, 6 authors, 2012-09-26

Re: [PATCH 05/10] mm, util: Use dup_user to duplicate user memory

From: Ezequiel Garcia <hidden>
Date: 2012-09-26 01:15:40
Also in: lkml

Hi Andrew,

On Tue, Sep 25, 2012 at 6:29 PM, Andrew Morton
[off-list ref] wrote:
On Sat,  8 Sep 2012 17:47:54 -0300
Ezequiel Garcia [off-list ref] wrote:
quoted
Previously the strndup_user allocation was being done through memdup_user,
and the caller was wrongly traced as being strndup_user
(the correct trace must report the caller of strndup_user).

This is a common problem: in order to get accurate callsite tracing,
a utils function can't allocate through another utils function,
but instead do the allocation himself (or inlined).

Here we fix this by creating an always inlined dup_user() function to
performed the real allocation and to be used by memdup_user and strndup_user.
This patch increases util.o's text size by 238 bytes.  A larger kernel
with a worsened cache footprint.

And we did this to get marginally improved tracing output?  This sounds
like a bad tradeoff to me.
Mmm, that's bad tradeoff indeed.
It's certainly odd since the patch shouldn't increase the text size
*that* much.
Is it too much to ask that you send your kernel config and gcc version.

My compilation (x86 kernel in gcc 4.7.1) shows a kernel less bloated:

$ readelf -s util-dup-user.o | grep dup_user
   161: 00001c10   108 FUNC    GLOBAL DEFAULT    1 memdup_user
   169: 00001df0   159 FUNC    GLOBAL DEFAULT    1 strndup_user
$ readelf -s util.o | grep dup_user
   161: 00001c10   108 FUNC    GLOBAL DEFAULT    1 memdup_user
   169: 00001df0    98 FUNC    GLOBAL DEFAULT    1 strndup_user

$ size util.o
   text	   data	    bss	    dec	    hex	filename
  18319	   2077	      0	  20396	   4fac	util.o
$ size util-dup-user.o
   text	   data	    bss	    dec	    hex	filename
  18367	   2077	      0	  20444	   4fdc	util-dup-user.o

Am I doing anything wrong?
If you still feel this is unnecessary bloatness, perhaps I could think of
something depending on CONFIG_TRACING (though I know
we all hate those nasty ifdefs).

Anyway, thanks for the review,
Ezequiel.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help