netdev - Re: [PATCH v7 2/3] netfilter: x_tables: Merge xt

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <6f276b2e-de9d-427e-a3a3-aac9ed340357@lunn.ch>
Date: Mon, 6 Jan 2025 15:22:53 +0100
From: Andrew Lunn <andrew@...n.ch>
To: egyszeregy@...email.hu
Cc: fw@...len.de, pablo@...filter.org, lorenzo@...nel.org,
	daniel@...earbox.net, leitao@...ian.org, amiculas@...co.com,
	kadlec@...filter.org, davem@...emloft.net, dsahern@...nel.org,
	edumazet@...gle.com, kuba@...nel.org, pabeni@...hat.com,
	horms@...nel.org, netfilter-devel@...r.kernel.org,
	coreteam@...filter.org, linux-kernel@...r.kernel.org,
	netdev@...r.kernel.org
Subject: Re: [PATCH v7 2/3] netfilter: x_tables: Merge xt_*.c files which has
 same name.

On Mon, Jan 06, 2025 at 12:31:56AM +0100, egyszeregy@...email.hu wrote:
> From: Benjamin Szőke <egyszeregy@...email.hu>
> 
> Merge xt_*.c source files, which has same upper and
> lower case name format. Combining these modules should
> provide some decent code size and memory savings.
> 
> Merge licenses, codes and adjuste Kconfig and Makefile
> for backwards-compatibility.
> 
> test-build:
> $ mkdir build
> $ wget -O ./build/.config https://pastebin.com/raw/teShg1sp
> $ make O=./build/ ARCH=x86 -j 16
> 
> x86_64-before:
> text    data     bss     dec     hex filename
>  716     432       0    1148     47c xt_dscp.o
> 1142     432       0    1574     626 xt_DSCP.o
>  593     224       0     817     331 xt_hl.o
>  934     224       0    1158     486 xt_HL.o
> 1099     120       0    1219     4c3 xt_rateest.o
> 2126     365       4    2495     9bf xt_RATEEST.o
>  747     224       0     971     3cb xt_tcpmss.o
> 2824     352       0    3176     c68 xt_TCPMSS.o
> total data: 2373
> 
> x86_64-after:
> text    data     bss     dec     hex filename
> 1709     848       0    2557     9fd xt_dscp.o
> 1352     448       0    1800     708 xt_hl.o
> 3075     481       4    3560     de8 xt_rateest.o
> 3423     576       0    3999     f9f xt_tcpmss.o
> total data: 2353

So you have saved 20 bytes in the data segment. A 0.8% reduction. If i
was developing this patchset, when i see this number, i would
immediately think, why am i bothering? It is not worth the effort for
the gains i'm getting. I'm also at risk of breaking the ABI and
userspace code, and 0.8% does not justify that risk.

This is why we ask for benchmarks. Benchmarks help to justify a
change. So far, there is no good justification for this.

    Andrew

---
pw-bot: cr