This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: [PATCH] Add support for Tilera TILE-Gx processor (part 2/2: gdb)
On 04/30/2012 07:51 PM, Jeff Kenton wrote:
> On 04/30/2012 02:43 PM, Pedro Alves wrote:
>> I meant to reply to this before, but it slipped...
>>
>> On 04/20/2012 04:23 PM, Jeff Kenton wrote:
>>
>> === gdb Summary ===
>>
>> # of expected passes 18181
>> # of unexpected failures 625
>>
>> Did you look at what is causing these hundreds of failures? Although there
>> are thousands of passes, such high failure rate is usually indicative of
>> something badly borked. You'd be surprised at how far in testsuite results
>> you can get with a gdb that manages to loads programs, but is a brick
>> at actually debugging live programs ...
>>
>>> # of expected failures 100
>>> # of known failures 59
>>> # of untested testcases 12
>>> # of unresolved testcases 4
>>> # of unsupported tests 118
>>> /gdb_tests/gdb/build/gdb/testsuite/../../gdb/gdb version 7.4.50.20120410 -nw -nx -data-directory /gdb_tests/gdb/build/gdb/testsuite/../data-directory
>>
>
> We use it on production systems, so lots of things work. Checking the failures is on my list of things to do. Is there a target failure threshhold we need to hit before our GDB is accepted, or can we submit what we have (with the fixes you specified) and work on the remaining test suite failures afterwards?
There's no threshold, and certainly zero fails are not expected. Not even the
most maintained ports get that far, unfortunately. So we'll accept the port anyway.
But I'd be happier if at least you had browsed the gdb.sum/gdb.log for obvious
issues, and at least (any) internal-errors that show up are given some
consideration.
--
Pedro Alves