Tryout - tryout latest toolchain, see gcc test/cov stats

Go To Last Post
12 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Tryout (http://tryout.senthilthecoder.co...) is a website I built for checking out the code the latest version of the toolchain (built nightly from source) generates. Figured it would be useful for people here - for compiling a snippet of code in a pinch without access to a dev environment, or just to see what the latest version of the toolchain does.

Just type in the code, specify the options and hit Compile. Note that I automatically add the -c flag (compile and assemble, no link) to whatever options you specify, and then invoke avr-objdump -S (generate diassembly with source annotations) on the generated object file.

For GCC devs - I also compile gcc with code coverage options and run gcov after running regression tests for mega and xmega - the formatted coverage results are published to http://tryout.senthilthecoder.co... . The mega regression test results are up at http://tryout.senthilthecoder.co... and xmega results at http://tryout.senthilthecoder.co... . (xmega tests still running ATM, so you'll be seeing yesterday's results until it finishes).

Internally, a cron job runs daily on a cheap VPS I own and invokes a bash script that
* checks out binutils, gcc and avr-libc from their respective source control URLs
* builds all three and copies to the tryout area
* checks out avrtest and builds it
* runs regression tests for an atmega device and an atxmega device
* copies the test results to the tryout area
* Runs gcov and then lcov to format the gcov output, and publishes to the tryout area.

Regards

Senthil

 

blog | website

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

So, what's the advantage of using your web site over just using GCC here?

The largest known prime number: 282589933-1

It's easy to stop breaking the 10th commandment! Break the 8th instead. 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:

So, what's the advantage of using your web site over just using GCC here?

Senthil is from the team that do the avr-gcc work for Atmel so this is an intriguing opportunity to see what's coming next.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Oh! I see! Try out the upcoming GCC. That's a different kettle of fish entirely, though I have to admit to barely using the current one. (GCC, that is, not kettle of fish.)

The largest known prime number: 282589933-1

It's easy to stop breaking the 10th commandment! Break the 8th instead. 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

In fact, I might have an idea. Beware of brats with ideas

The largest known prime number: 282589933-1

It's easy to stop breaking the 10th commandment! Break the 8th instead. 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
Beware of brats with ideas

Next time I see any ideas coming out of my brats, I'll take them back to the store. :D

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

If your "fried sausage" has ideas, they are definitely NO GOOD.

The largest known prime number: 282589933-1

It's easy to stop breaking the 10th commandment! Break the 8th instead. 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

If you are running these tests on a regular basis, it might be a good idea to post the summaries (.sum) to the GCC testresults mailing list:

http://gcc.gnu.org/ml/gcc-testre...

That would show to the GCC maintainers that the avr port is not (completely) dead...

Appreciated results would be for 4.7 and 4.8 HEAD, and for trunk (future 4.9).

avrfreaks does not support Opera. Profile inactive.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

First, nice job, Senthil!

I just tried the "online compiler", and have a request - could you please instead of performing -c and objdump, rather, compile with -S and output that?

The reason for this is, that in the unlinked binary's disassembly, all the local jumps are unresolved and the symbols are missing, so it's usability is severely limited...

Thanks,

Jan

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Thanks Jan. I've changed it to just compile now - no assembly imvolved.

I'd gone with disassembling object files because the specific case I was interested in involved the assembler/disassembler (generation/reading of debug_line data). But yes, you're right in that nearly everything else looks much better with -S.

Regards

Senthil

 

blog | website

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Thanks.

JW

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

SprinterSB wrote:
If you are running these tests on a regular basis, it might be a good idea to post the summaries (.sum) to the GCC testresults mailing list:

Starting yesterday, mega and xmega test results for gcc trunk are being sent to the mailing list.

I'd set this up much earlier, but trunk was broken for AVR for a while.

Regards

Senthil

 

blog | website