Re: [avr-gcc-list] More results from the testsuite with avrtest

[Paulo Marques]

Andy Hutchinson wrote:
> Paulo,

Hi, Andy.

> please report bug with PR27364 testcase.
>
> I found that problem is somewhere in combine phase. Here it is trying to 
> stick  together pairs of instruction to make another. It fails but that 
> is where load of constant dissappears.
>
> So if you post bug, I can add further info.

Ok. I'll post it tonight, then.

> Andy
>
> PS running your simulator now. Found bug in AVRORA - dunno where but it 
> failed some testcases.

Tracking down a long sequence of operations that end in a bug can be 
really tricky (and I'm sure you've been there, too).

But if avrora fails a testcase that avrtest passes, it doesn't 
necessarily mean that there is a bug with avrora. It is likely that the 
bug is in avrtest instead.

This is still a concern of mine: when a test fails, it is "relatively" 
easy to find the culprit: gcc or avrtest. But if it passes, I won't even 
go and look for it, so I might get a false positive.

Your tests with avrora were really helpful in testing both our setups.

> Your's is same speed. Sadly Cygwin does not like spawn.

Not that bad, considering that avrtest wasn't written with speed in mind :)

However, as I said in another mail, the speed of simulator actually 
counts very little to the overall speed of the testsuite. Even a 
simulator that took twice the time to execute than avrtest, would only 
increase the total time by 5%.

> Hopefully, our numbers will match!

"Hopefully" was a wise choice of words ;)

Thanks for your help,

--
Paulo Marques
Software Development Department - Grupo PIE, S.A.
Phone: +351 252 290600, Fax: +351 252 290601
Web: www.grupopie.com

"As far as we know, our computer has never had an undetected error."
Weisert