ffsend segfaults on upload

[Newchair2644]

When attempting to upload a file using ffsend upload, ffsend segfaults. Sample output:

$ ffsend upload Pictures/Saved/oh-yeah-kool-aid-man-3596919834.gif 
Upload complete  
Segmentation fault

Same result with ffsend debug:

$ ffsend debug
Segmentation fault

I am using void linux with the ffsend package provided in the repos, which is up to date 0.2.76.

[FireMasterK]

I've noticed the same on Alpine Linux too, and have opened an issue on aports too: https://gitlab.alpinelinux.org/alpine/aports/-/issues/15158

[uzza1hossain]

same issue on m1 pro.

[Saeden]

Same issue on Fedora 39, using the cargo install version of ffsend. The new static binary does not have this issue.

[noyez]

just and fyi, i have observed this too from the version installed via cargo install ffsend, but if i compile directly from master i do not see a crash.

[jannic]

Stacktrace of the crash:

(gdb) info stack
#0  0x000055ded771cf66 in prettytable::TableSlice::get_all_column_width ()
#1  0x000055ded771d382 in prettytable::TableSlice::print_tty ()
#2  0x000055ded76c5180 in ffsend::action::upload::Upload::invoke ()
#3  0x000055ded76d32a0 in ffsend::main ()
#4  0x000055ded76b5ed3 in std::sys_common::backtrace::__rust_begin_short_backtrace ()
#5  0x000055ded76cd639 in std::rt::lang_start::{{closure}} ()
#6  0x000055ded7b09e90 in std::rt::lang_start_internal ()
#7  0x000055ded76d4365 in main ()

This makes me guess that this is caused by https://github.com/phsym/prettytable-rs/issues/145.

That would fit the observation that the segfault doesn't happen with ffsend built from git: That one uses prettytable 0.10.0, which contains a fix for the linked issue.

[jannic]

Oh, there's also https://github.com/timvisee/ffsend/pull/154 and https://github.com/timvisee/ffsend/issues/153, so this is just a duplicate of a known and fixed issue.

[phispi]

Confirmed on my system as well: Linux 6.1.0-22-amd64 #1 SMP PREEMPT_DYNAMIC Debian 6.1.94-1 (2024-06-21) x86_64 GNU/Linux.

Installation with cargo install ffsend was leading to a segmentation fault after upload, whereas using the compiled version was running smoothly with the identical input data.