Mr. Waldek,
Could you please publish your runner script so I could run properly test suite despite the presence of PAX process killer on my Linux kernel?
IMHO these half-done tests are of no use to anybody, so please if you would be so kind to grant my request.
Regards, Mirsad
"Be conservative in what you do, be liberal in what you accept from others." -- Jon Postel
Mirsad Todorovac wrote:
Mr. Waldek,
Could you please publish your runner script so I could run properly test suite despite the presence of PAX process killer on my Linux kernel?
Sorry, ATM I can not find it. I will be probably easier to re-create it. Anyway what I had would require work to bring it to release quality (probably much more than I spent on creating my internal version).
IMHO these half-done tests are of no use to anybody, so please if you would be so kind to grant my request.
Actually I think that results of your tests are pretty clear: the failing tests are the same as on the other platforms (and they are not very serious). AFAIU the test which produced killed message did work when you have disabled kernel executable stack protection. Which means that from gpc point of view you got expected positive result. I would say that the only remaining thing is to find minimal set of flags needed to run programs using trampolines.
Runner script would be usefull for other folks to be able do tests in fully automatic way, but IMHO manual way (run all tests with standard setting and re-run the ones failing with Killing message using reduced protection) is as valid as automatic one, just tedious.
Unfortunatly I am now using most of my free time on other projects so can spent only little time on gpc. However I will get to this sooner or later (unless sombody will do it earlier).
On Tue, 27 Feb 2007, Waldek Hebisch wrote:
Mirsad Todorovac wrote:
Could you please publish your runner script so I could run properly test suite despite the presence of PAX process killer on my Linux kernel?
Sorry, ATM I can not find it. I will be probably easier to re-create it. Anyway what I had would require work to bring it to release quality (probably much more than I spent on creating my internal version).
Sorry to hear this. AFAIAC, any working script would do. It does not have to be release quality ATM. I am pretty familiar with situation, having not met Frank's standards with other programmed stuff ;-)
IMHO these half-done tests are of no use to anybody, so please if you would be so kind to grant my request.
Actually I think that results of your tests are pretty clear: the failing tests are the same as on the other platforms (and they are not very serious). AFAIU the test which produced killed message did work when you have disabled kernel executable stack protection. Which means that from gpc point of view you got expected positive result. I would say that the only remaining thing is to find minimal set of flags needed to run programs using trampolines.
I am not very familiar with tranmpolines, and the way they are emulated. If this is an issue of importance for GPC testing (which it obviously is), then I need some background information.
Runner script would be usefull for other folks to be able do tests in fully automatic way, but IMHO manual way (run all tests with standard setting and re-run the ones failing with Killing message using reduced protection) is as valid as automatic one, just tedious.
Exactly.
Unfortunatly I am now using most of my free time on other projects so can spent only little time on gpc. However I will get to this sooner or later (unless sombody will do it earlier).
If the problem is taht you have found and old version and just that the version you have found in your archive is too untidy for genpop, I would be happy just if it completes the task.
Regards, Mirsad
-
Mirsad Todorovac -
Waldek Hebisch