Message-ID: <1432063162.2870.199.camel@perches.com>
Subject: Re: mod_devicetable: Make dmi_strmatch.substr const char *
From: Joe Perches <joe@perches.com>
To: One Thousand Gnomes <gnomes@lxorguk.ukuu.org.uk>
Cc: David Woodhouse <dwmw2@infradead.org>,
        Rusty Russell <rusty@rustcorp.com.au>,
        LKML <linux-kernel@vger.kernel.org>,
        Quentin Casasnovas <quentin.casasnovas@oracle.com>
Date: Tue, 19 May 2015 12:19:22 -0700
In-Reply-To: <20150519165630.169c64d3@lxorguk.ukuu.org.uk>
References: <1431994055.2870.102.camel@perches.com>
	 <1432018018.3277.22.camel@infradead.org>
	 <20150519165630.169c64d3@lxorguk.ukuu.org.uk>
Content-Type: text/plain; charset="ISO-8859-1"
Mime-Version: 1.0
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 2871
Lines: 81

On Tue, 2015-05-19 at 16:56 +0100, One Thousand Gnomes wrote:
> On Tue, 19 May 2015 07:46:58 +0100 David Woodhouse <dwmw2@infradead.org> wrote:
> > On Mon, 2015-05-18 at 17:07 -0700, Joe Perches wrote:
> > > changed dmi_strmatch.substr from char * to char[79];
> > > 
> > > Changing it back to const char * would shrink an x86-64
> > > defconfig more than 100KB.
> > > 
> > > $ size vmlinux.old vmlinux.new
> > >     text    data     bss      dec    hex filename
> > > 11941725 1825624 1085440 14852789 e2a2b5 vmlinux.old
> > > 11921172 1730648 1085440 14737260 e0df6c vmlinux.new
> 
> What percentage of those are __initdata ?

old: 17 .init.data    000876b0  ffffffff81f9e000  0000000001f9e000  0139e000  2**12
new: 17 .init.data    000711b0  ffffffff81f9d000  0000000001f9d000  0139d000  2**12

.init.data: 0x876b0 - 0x711b0 = 0x16500 (91392)
   vmlinux: 14852789 - 14737260 = 115529

so there's ~25KB delta that's not .init.data.

The longest DMI_MATCH substr I found was 40 chars, so
there's some value in reducing the substr size of 79
to something shorter like 47 to reduce 80*4=320 to
48*4=192 per use.

$ grep-2.5.4 -rP  --include=*.[ch] "\bDMI_(?:EXACT_)?MATCH\s*\(\s*\w+\s*,\s*\"[^\"]*" *| \
  grep -oh '"[^"]*"' | \
  sed 's/"//g' | sort | uniq | \
  awk '{print length($0), "	", $0}'| \
  sort -rn | head -10
39     Matsushita Electric Industrial Co.,Ltd.
35     ASUS PR-DLS ACPI BIOS Revision 1010
34     Micro-Star International Co., Ltd.
34     Award Software International, Inc.
34     900X3C/900X3D/900X3E/900X4C/900X4D
34     370R4E/370R4V/370R5E/3570RE/370R5V
33     MICRO-STAR INTERNATIONAL CO., LTD
32     MV85010A.86A.0016.P07.0201251536
32     MO81010A.86A.0008.P04.0004170800
32     ASUS A7V ACPI BIOS Revision 1007

Changing substr from 79 to 47 reduces .init.data a bit
(different -next version)

old: 17 .init.data    00081718  ffffffff81f84000  0000000001f84000  01384000  2**12
new: 17 .init.data    00076598  ffffffff81f84000  0000000001f84000  01384000  2**12

.init.data: 0x81718 - 0x76598 = 0xb180 (45440)

Unless/until modpost is updated, maybe this patch is OK:
---
 include/linux/mod_devicetable.h | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/include/linux/mod_devicetable.h b/include/linux/mod_devicetable.h
index 3bfd567..279f1be 100644
--- a/include/linux/mod_devicetable.h
+++ b/include/linux/mod_devicetable.h
@@ -462,7 +462,11 @@ enum dmi_field {
 struct dmi_strmatch {
 	unsigned char slot:7;
 	unsigned char exact_match:1;
-	char substr[79];
+#ifdef CONFIG_MODULES
+	char substr[47];
+#else
+	const char *substr;
+#endif
 };
 
 struct dmi_system_id {


--
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/