From 5dd846b73784f20532281538877ef6fe6b8a4cfa Mon Sep 17 00:00:00 2001 From: "https://www.google.com/accounts/o8/id?id=AItOawma2Ssl7yrV7NQ6zPklPtlymc2-WC_bt98" Date: Wed, 10 Jul 2013 19:32:34 +0200 Subject: Wrote my 3rd week report. --- user/NlightNFotis.txt | 15 ++++++++++++++- 1 file changed, 14 insertions(+), 1 deletion(-) (limited to 'user') diff --git a/user/NlightNFotis.txt b/user/NlightNFotis.txt index a2fb3c74..914d7820 100644 --- a/user/NlightNFotis.txt +++ b/user/NlightNFotis.txt @@ -79,5 +79,18 @@ Projects: - Started hacking on the code, attempting to build the GCC codebase on the Hurd, Debian x64 (host) and XUbuntu x86 (another guest) and have not met any success in any of them. - My builds fail on each of them with linker errors. Fortunately, the errors are reproducible across the different platforms, and I have documented them here https://gist.github.com/NlightNFotis/5896188#file-build_results - - Next week schedule: + - Next week schedule: - I am trying to overcome the obstacles I am facing with building GCC. I am currently investigating the issue, and will report back as soon as I come up with a solution. + + - July 2 - July 9 + - Continued attempting to build GCC on the Hurd. Managed to get past the linker errors pinpointed at the last report. + - After getting past those errors, some other errors showed up. Fortunately they were tracked down with help from Hurd developers in the #hurd channel. The specific problem was that compilations seemed to completely freeze at some point. The reason this happened was that the Hurd image I was using wan't large enough for gcc to compile successfully (it was only 3GiB). I manually attempted to resize the qemu image, which was fairly easy to do so, however I had to employ several courses of action inside the VM in order to get it to recognize the extra disc space that was added. + - After dealing with this issue too, I re attempted the compilation phase but this time stopped with compilation errors. The compiler log can be found here: https://www.dropbox.com/s/u5nx5ykebdx4sk5/build_results.txt + - Having discussed that issue with my mentor Thomas Schwinge in the weekly meeting on July 8th I was told that there was a patch employed that was supposed to fix that issue. Seeing as in my case it was not so, it was suggested to me that I check the configuration and the state of my GCC repository inside the Hurd, make sure it's on the correct working branch, and that it had the latest commits from that branch pulled. Seeing as all the preconditions were met, I was also told to mount the Hurd image under my Linux host box, and get diff results for the Host and Vm repository of GCC. I have got 2 diffs, both of which don't have any relationship with the code that fails the build. For reference, my diff is here: https://www.dropbox.com/s/ilgc3skmhst7lpv/diffs_in_git.txt + - Even though there didn't seem to be any issue with the integrity of the GCC repository under the Hurd, I still decided to clear the working directory, and recheckout the branch. Having done so, the results are the same with the original compilation errors (indicating that there must be an issue with the code itself. + + - Next week actions: + - I am attempting to fix the issues that I am facing now (the ones with init_context and fix_context failing the compilation) + - Meanwhile I want to attempt a compilation of GCC with the -w flag passed to make, so that warnings do not fail the compilation (the compilation fails because the -Werror option is passed to the compiler, and both init_context and fix_context produce a warning, that is elevated to an error due to that option) and see if it manages to build itself under the hurd and pass (at least some of its tests) + + -- cgit v1.2.3