Re: GRUB 2.00 build and flex-2.5.37

["Vladimir 'φ-coder/phcoder' Serbinenko" <phcoder@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 1688 bytes --]

On 02.01.2013 11:22, Colin Watson wrote:

> On Sun, Aug 05, 2012 at 05:20:46PM -0500, Bruce Dubbs wrote:
>> The latest flex makes the GRUB-2.00 build fail.  I can work around
>> the issues with --disable-werror, but grub-core/script/yylex.l
>> causes two warnings that do not work with -Werror.
>>
>> ./grub-core/script/yylex.l: At top level:
>> grub_script.yy.c:2351:13: error: 'yy_fatal_error' defined but not used
>>
>> This can be fixed by removing  #define YY_FATAL_ERROR(msg) in yylex.l.
> 
> That would be incorrect; the default implementation of yy_fatal_error
> uses fprintf, which is not valid in parts of GRUB that need to run
> freestanding.  More correct would probably be to insert an artificial
> reference to yy_fatal_error, since flex offers no way to suppress its
> definition.
> 

It's ok to make such warnings non-breaking by corresponding #pragma. In
fact we already have:
#pragma GCC diagnostic ignored "-Wunused-function"
So the report is obviously for some older version.

>> grub_script.yy.c: In function 'yy_scan_bytes':
>> grub_script.yy.c:2296:17: error: comparison between signed and
>> unsigned integer expressions [-Werror=sign-compare]
>>
>> This one is a little more difficult because it is generated by flex.
>> I don't know how to fix this except with a sed after
>> grub_script.yy.c is generated by flex.
> 
> I tried to reproduce this and the other warning with flex 2.5.37 and GCC
> 4.7.2 and couldn't, so I'm reluctant to make a change that I can't test.
> Can you still reproduce this?  If so, could you attach your generated
> grub_script.yy.c?
> 



-- 
Regards
Vladimir 'φ-coder/phcoder' Serbinenko


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 294 bytes --]