From: Denis Vlasenko <vda@ilport.com.ua>
To: pmarques@grupopie.com, kai.germaschewski@gmx.de, kai@germaschewski.name
Cc: linux-kernel <linux-kernel@vger.kernel.org>, akpm@osdl.org
Subject: [PATCH] do not save thousands of useless symbols in KALLSYMS kernels
Date: Thu, 11 Aug 2005 14:03:39 +0300 [thread overview]
Message-ID: <200508111403.39708.vda@ilport.com.ua> (raw)
[-- Attachment #1: Type: text/plain, Size: 620 bytes --]
Sample of my kernel's mostly useless symbols
(starting_with:# of symbols):
__func__: 624
__vendorstr_: 1760
__pci_fixup_PCI_: 116
__ksymtab_: 2597
__kstrtab_: 2597
__kcrctab_: 2597
__initcall_: 236
__devicestr_: 4686
__devices_: 1760
Total: 16973
Lines in System.map: 39735
Excluding them from in-kernel symbol table saves ~300kb:
text data bss dec hex filename
4337710 1054414 259296 5651420 563bdc vmlinux.carrier1 - w/o KALLSYMS
4342068 1296046 259296 5897410 59fcc2 vmlinux - with KALLSYMS+patch
4341948 1607634 259296 6208878 5ebd6e vmlinux.carrier - with KALLSYMS
^^^^^^^
--
vda
[-- Attachment #2: kallsyms.c.diff --]
[-- Type: text/x-diff, Size: 1685 bytes --]
--- linux-2.6.12.org/scripts/kallsyms.c.org Sun Jun 19 16:11:06 2005
+++ linux-2.6.12.org/scripts/kallsyms.c Wed Aug 10 23:52:06 2005
@@ -36,7 +36,7 @@
/* we use only a subset of the complete symbol table to gather the token count,
* to speed up compression, at the expense of a little compression ratio */
-#define WORKING_SET 1024
+#define WORKING_SET (1024*4)
/* first find the best token only on the list of tokens that would profit more
* than GOOD_BAD_THRESHOLD. Only if this list is empty go to the "bad" list.
@@ -102,11 +102,34 @@ usage(void)
* This ignores the intensely annoying "mapping symbols" found
* in ARM ELF files: $a, $t and $d.
*/
+/*
+__func__: 624
+__vendorstr_: 1760
+__pci_fixup_PCI_: 116
+__ksymtab_: 2597
+__kstrtab_: 2597
+__kcrctab_: 2597
+__initcall_: 236
+__devicestr_: 4686
+__devices_: 1760
+Total: 16973
+Lines in System.map: 39735
+*/
static inline int
is_arm_mapping_symbol(const char *str)
{
- return str[0] == '$' && strchr("atd", str[1])
- && (str[2] == '\0' || str[2] == '.');
+ return (str[0] == '$' && strchr("atd", str[1])
+ && (str[2] == '\0' || str[2] == '.')
+ )
+ || (0 == strncmp(str, "__ksymtab_", sizeof("__ksymtab_")-1))
+ || (0 == strncmp(str, "__kstrtab_", sizeof("__kstrtab_")-1))
+ || (0 == strncmp(str, "__kcrctab_", sizeof("__kcrctab_")-1))
+ || (0 == strncmp(str, "__devices_", sizeof("__devices_")-1))
+ || (0 == strncmp(str, "__devicestr_", sizeof("__devicestr_")-1))
+ || (0 == strncmp(str, "__vendorstr_", sizeof("__vendorstr_")-1))
+ || (0 == strncmp(str, "__pci_fixup_PCI_", sizeof("__pci_fixup_PCI_")-1))
+ || (0 == strncmp(str, "__func__", sizeof("__func__")-1))
+ ;
}
static int
next reply other threads:[~2005-08-11 11:04 UTC|newest]
Thread overview: 3+ messages / expand[flat|nested] mbox.gz Atom feed top
2005-08-11 11:03 Denis Vlasenko [this message]
2005-08-11 11:16 ` Paulo Marques
2005-08-11 11:38 ` Denis Vlasenko
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=200508111403.39708.vda@ilport.com.ua \
--to=vda@ilport.com.ua \
--cc=akpm@osdl.org \
--cc=kai.germaschewski@gmx.de \
--cc=kai@germaschewski.name \
--cc=linux-kernel@vger.kernel.org \
--cc=pmarques@grupopie.com \
/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
all inboxes | Powered by JetHome®