This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: (bootstrap-debug) introduce make bootstrap-debug


On Aug 16, 2007, "Andrew Pinski" <pinskia@gmail.com> wrote:

> On 8/16/07, Alexandre Oliva <aoliva@redhat.com> wrote:
>> This patch enables us to test whether -g0 and -g2 do indeed generate
>> identical code, as intended.  With this patch, stage2 is built with
>> -g0 and stage3 with -g2.  I suppose I could turn these into Makefile
>> macros for additional flexibility, and leave stage3's flags alone,
>> while setting stage2's additional -g flag to -g or -g0 depending on
>> whether BOOT_CFLAGS contains -g, but I'll leave that for a separate
>> patch.

> How about just adding a testsuite instead of changing bootstrap?

A testsuite that runs bootstrap with different flags for stage2 and
stage3?  That's doable, but that would make the test cycle quite long,
would it not?  In fact, using the newly-introduced make
bootstrap-debug might actually *shorten* the test cycle, while at the
same time giving us broader coverage.

> Plus there is already a testcase like this already, see
> gcc.c-torture/unsorted/dump-noaddr.[cx] .

AFAICT this only tests that the tree dumps are the same.  This is not
quite the same as verifying that we generate the exact same code with
or without -g, which is what this new bootstrap target is about.

-- 
Alexandre Oliva         http://www.lsd.ic.unicamp.br/~oliva/
FSF Latin America Board Member         http://www.fsfla.org/
Red Hat Compiler Engineer   aoliva@{redhat.com, gcc.gnu.org}
Free Software Evangelist  oliva@{lsd.ic.unicamp.br, gnu.org}


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]