Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Date: Thu, 2 Mar 2017 16:42:11 -0800
From: Reinoud Koornstra <>
To: Rich Felker <>
Subject: Re: getting rid of some warnings

On Thu, Mar 2, 2017 at 4:27 PM, Rich Felker <> wrote:
> On Thu, Mar 02, 2017 at 02:37:19PM -0800, Reinoud Koornstra wrote:
>> Hello Rich,
>> I did compile against musl using the musl-gcc build from your scripts.
>> ->MUSL-GCC/bin/x86_64-linux-musl-gcc -nostdinc
>> -I/users/koorstra/MUSL-GCC/x86_64-linux-musl/include/ -Wall -g -o
>> split_buff split_buff.c
>> In file included from
>> /users/koorstra/MUSL-GCC/x86_64-linux-musl/include/sys/types.h:70:0,
>>                  from split_buff.c:4:
>> /users/koorstra/MUSL-GCC/x86_64-linux-musl/include/endian.h: In
>> function '__bswap32':
>> /users/koorstra/MUSL-GCC/x86_64-linux-musl/include/endian.h:32:25:
>> warning: suggest parentheses around arithmetic in operand of '|'
>> [-Wparentheses]
>>   return __x>>24 | __x>>8&0xff00 | __x<<8&0xff0000 | __x<<24;
>>                    ~~~~~~^~~~~~~
> This happened because you're not just using the musl-gcc wrapper
> script, but passing -nostdinc and -I to get the system headers.
> Instead either configure for the right include path so you don't need
> to do this, or use -isystem instead of -I.

Ah, I couldn't find that script, that why I used the currently options.
Where is that script located? I did a search on the fs for musl-gcc,
but no results.
Also, I tried your suggestion: ->MUSL-GCC/bin/x86_64-linux-musl-gcc
-isystem -Wall -g -o split_buff split_buff.c
and that did compile without any warnings. The code i'm compiling is
own written code that doesn't yet have a config script.

>> /users/koorstra/MUSL-GCC/x86_64-linux-musl/include/endian.h:32:41:
>> warning: suggest parentheses around arithmetic in operand of '|'
>> [-Wparentheses]
>>   return __x>>24 | __x>>8&0xff00 | __x<<8&0xff0000 | __x<<24;
>>                                    ~~~~~~^~~~~~~~~
>> /users/koorstra/MUSL-GCC/x86_64-linux-musl/include/endian.h: In
>> function '__bswap64':
>> /users/koorstra/MUSL-GCC/x86_64-linux-musl/include/endian.h:37:23:
>> warning: suggest parentheses around '+' inside  <<' [-Wparentheses]
>>   return __bswap32(__x)+0ULL<<32 | __bswap32(__x>>32);
>>          ~~~~~~~~~~~~~~^~~~~
>>  koorstra@...pefer ~
>> ->readelf -d split_buff | grep NEEDED
>>  0x0000000000000001 (NEEDED)             Shared library: []
>>  koorstra@...pefer ~
>> ->./split_buff
>> -bash: ./split_buff: /lib/ bad ELF interpreter: No
>> such file or directory
>> How can I get rid of these warnings?
>> Also, I guess I need to set the default library paths different?
>> Any advice for this?
>> Thanks,
> If you dynamic link, the dynamic linker ("ELF interpreter") has to be
> present at runtime. Your options are:
> 1. Install it in /lib as above.
> 2. Static link.
> 3. Explicitly invoke it from another location to run the program:
>    /path/to/ ./split_buff ...
> 4. Configure musl with a different --syslibdir. Note that if you do
>    this, programs you link will only work with ld-musl installed in
>    the location you have it on your system, and won't be portable to
>    standard musl-based systems.
> I would recommend option 1 if it's your own system and you can, option
> 2 or 3 for something you need to ship (you can wrap option 3 with a
> script), and option 4 if this is just for testing or personal use (not
> shipping) on a system you don't have root on.

Thanks, the following worked for me too:
./MUSL-GCC/x86_64-linux-musl/lib/ split_buff
Other issue is that is a symlink to /lib/
........ which isn't the right one.
This can be remedied with changing the -syslibdir as you mentioned correct?


> Rich

Powered by blists - more mailing lists

Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.