Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

on archlinux the gcc 7.1 build fails #6

Closed
ChristophHaag opened this issue Aug 25, 2017 · 5 comments
Closed

on archlinux the gcc 7.1 build fails #6

ChristophHaag opened this issue Aug 25, 2017 · 5 comments

Comments

@ChristophHaag
Copy link

But not because of a segfault

./md-unwind-support.h: In function 'x86_64_fallback_frame_state':
./md-unwind-support.h:65:47: error: dereferencing pointer to incomplete type 'struct ucontext'
       sc = (struct sigcontext *) (void *) &uc_->uc_mcontext;
                                               ^~

Sounds like the gcc 7.2 sources should compile: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=81712

@BuzzBumbleBee
Copy link

@ChristophHaag see my PR, seems to work fine on arch (as of yesterday) for me

@suaefar
Copy link
Owner

suaefar commented Aug 27, 2017

Sorry, but I had to remove the support for Arch. I suppose Arch users know what to do to get the script running, or how to reproduce the buggy behaviour for that matter. The problem with Arch is, that I can't rely on a standard version of, e.g., gcc. The reference implementation will be based on Ubuntu.

@suaefar suaefar closed this as completed Aug 27, 2017
@ttbek
Copy link

ttbek commented Sep 21, 2017

I updated my Arch yesterday and got this error running the script this morning, if it is fixed upstream it may not have filtered down into Arch yet. Before my update it was building on Arch quite happily. I get that you only want to support a single OS, but it will be good for people to know they may encounter this issue, I will update the AMD forum thread. I'm not quite clear on the referenced link, they updated gcc or glib? Looks like gcc to me, so they decided glib is behaving correctly. I did switch to gcc-multilib (needed to make some 32 bit static binaries) though, maybe the change didn't reach the multilib version yet.
I wouldn't assume Arch users are necessarily that much more advanced.... I'm actually running Arch because when I was installing my OS Ubuntu had significant problems running on the Gigabyte X370 K7 board.... and turns out Gigabyte doesn't give a hoot.

@Oxalin
Copy link
Contributor

Oxalin commented Oct 10, 2017

Not related to rolling release distributions. Problem comes with glib 2.26 under gcc 7.1.0 known bug and fixed, but not backported), which is fixed under gcc 7.2: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=81712

Updating the script to download gcc 7.2.0 (.tar.xz please) should fix the problem.

Oxalin added a commit to Oxalin/ryzen-test that referenced this issue Oct 10, 2017
…to xz archive since it offers a better compression, so shorter to download.

Fixes real cause of issue suaefar#6.

Signed-off-by: Alexandre Demers <alexandre.f.demers@gmail.com>
suaefar pushed a commit that referenced this issue Oct 24, 2017
…rovements

* Don't fail if folder already exists. We may be trying to launch the script a second time for some reasons.

Signed-off-by: Alexandre Demers <alexandre.f.demers@gmail.com>

* Use GCC 7.2 source code, since 7.1 is broken against glib 2.26. Move to xz archive since it offers a better compression, so shorter to download.

Fixes real cause of issue #6.

Signed-off-by: Alexandre Demers <alexandre.f.demers@gmail.com>

* Warn if we have detected a rolling release distribution based on ArchLinux
or if this is an unsupported distribution.

User is on his own.

Signed-off-by: Alexandre Demers <alexandre.f.demers@gmail.com>

* Try to create the ramdisk on mountpoint only if it doesn't already exists.

Signed-off-by: Alexandre Demers <alexandre.f.demers@gmail.com>

* Add a variable so the GCC archive to be downloaded only need to be changed
once for everywhere. Prevent downloading more than once the file. Continue
to download if it had been interrupted.

Signed-off-by: Alexandre Demers <alexandre.f.demers@gmail.com>

* Fix typo in comment.

Signed-off-by: Alexandre Demers <alexandre.f.demers@gmail.com>

* Removed superfluous "="

It shouldn't have gotten there in the first place.
@Cimbali
Copy link

Cimbali commented Jan 31, 2018

FYI the problem is just that struct ucontext doesn't exist anymore in glibc 2.26.

As a workaround to compile gcc code with glibc 2.26+ you can change struct ucontext to ucontext_t in the gcc source.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

6 participants