linux-kernel - Re: 'Re: [PATCH] drivers/android: remove redundant ret variable'

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] [day] [month] [year] [list]

Message-ID: <YdVuZRps9HP0Udmc@kroah.com>
Date:   Wed, 5 Jan 2022 11:09:41 +0100
From:   Greg KH <gregkh@...uxfoundation.org>
To:     cgel.zte@...il.com
Cc:     arve@...roid.com, chi.minghao@....com.cn, christian@...uner.io,
        hridya@...gle.com, joel@...lfernandes.org,
        linux-kernel@...r.kernel.org, maco@...roid.com, surenb@...gle.com,
        tkjos@...roid.com, zealci@....com.cn
Subject: Re: 'Re: [PATCH] drivers/android: remove redundant ret variable'

On Wed, Jan 05, 2022 at 09:19:42AM +0000, cgel.zte@...il.com wrote:
> I found that there are indeed some optimizations
> after removing the redundancy.

You stripped off all context, making this impossible to remember what
you are responding to :(

> 
> For example:
> 
> int foo(int a, int b)
> {
>         int ret;
>         ret = a>b ? a: b;
>         return ret;
> }
> int main()
> {
>         int val;
>         val = foo(7,2);
>         return 0;
> }
> 
> and
> 
> int foo(int a, int b)
> {
>         return a>b ? a: b;
> }
> int main()
> {
>         int val;
>         val = foo(7,2);
>         return 0;
> }
> 
> The corresponding disassembly code is as follows：
> 
>     1129:       f3 0f 1e fa             endbr64 
>     112d:       55                      push   %rbp
>     112e:       48 89 e5                mov    %rsp,%rbp
>     1131:       89 7d fc                mov    %edi,-0x4(%rbp)
>     1134:       89 75 f8                mov    %esi,-0x8(%rbp)
>     1137:       8b 45 fc                mov    -0x4(%rbp),%eax
>     113a:       39 45 f8                cmp    %eax,-0x8(%rbp)
>     113d:       0f 4d 45 f8             cmovge -0x8(%rbp),%eax
>     1141:       5d                      pop    %rbp
>     1142:       c3                      retq
> 
> and
> 
>     1129:       f3 0f 1e fa             endbr64 
>     112d:       55                      push   %rbp
>     112e:       48 89 e5                mov    %rsp,%rbp
>     1131:       89 7d ec                mov    %edi,-0x14(%rbp)
>     1134:       89 75 e8                mov    %esi,-0x18(%rbp)
>     1137:       8b 45 ec                mov    -0x14(%rbp),%eax
>     113a:       39 45 e8                cmp    %eax,-0x18(%rbp)
>     113d:       0f 4d 45 e8             cmovge -0x18(%rbp),%eax
>     1141:       89 45 fc                mov    %eax,-0x4(%rbp)
>     1144:       8b 45 fc                mov    -0x4(%rbp),%eax
>     1147:       5d                      pop    %rbp
>     1148:       c3                      retq
> 
> After removing the redundancy, the compiler does
> have some optimizations

As I said on your other email, look at the kernel built files please,
they should be much different from your tiny example above.

thanks,

greg k-h