IRC channel logs

<dsmith>ekaitz, Just curious, do the same ones fail the same way every time? Or are they "intermittent"?

<ekaitz>dsmith: i just tried once but I think all of the ones that use threads fail

<ekaitz>oh! concurrent web-server passed!

<ekaitz>dsmith: ok, now the failures are different

<ekaitz>and concurrent web-server failed

<ekaitz>so yeah, they are intermittent

<dsmith>Yeah, that makes things a bit tougher

<ekaitz>do you think?

<ekaitz>maybe it's just some memory barrier that's wrong

<ekaitz>we might be lucky

<ekaitz>ACTION is a fake optimist

<dsmith>There is a linux setting that stops address space randomization.

<dsmith>Might be useful to set that and see how things behave

<dsmith>(there was an arm jit bug that depended on the address)

<ekaitz>I need to collect some courage to actually try to fix this

<ekaitz>i'll leave it for some sleepless night

<ekaitz>ACTION means *any* night :)

<dsmith>And I just can *not* remember how to change that address space thingy

<ekaitz>dsmith: don't worry, just ping me if you remember

<dsmith>Looks like it might be `setarch --addr-no-randomize {program}`

<ekaitz>dsmith: great! I have to change things carefully to test with it

<ekaitz>but i have that written down for later

<ekaitz>thank you

<dsmith>np. I'd like to help, but don't have the bandwidth at this time

<ekaitz>me neither :)

<old>setarch `uname -m` --addr-no-randomize bash is your friend

<old>ekaitz: if you have a backtrace that would be ince

<old>you could get a coredump and make a backtrace out of it

<ekaitz>gimme a sec and i'll try

<ekaitz>okay i found my bug

<ekaitz>now context switch to fibers

<old>so, what was the bug?

<ekaitz>ah it was some GNU Mes bug, a different thing

<ekaitz>i had a very interesting buffer-overflow

2026-03-23.log