#448589 scp: Segfaults when recursivly transfering files from a La Fonera (running dropbear) #448589
- Package:
- openssh-client
- Source:
- openssh
- Description:
- secure shell (SSH) client, for secure access to remote machines
- Submitter:
- Tobias Frost
- Date:
- 2014-08-02 08:09:19 UTC
- Severity:
- normal
Issuing e.g:
scp -r root@fonera:/sbin .
root@fonera's password:
Segmentation fault
heres the last lines of the strace, reading of the password removed.
stat64(".", {st_mode=S_IFDIR|0755, st_size=4096, ...}) = 0
read(7, root@fonera's password:
(some lines removed)
stat64("./sbin", {st_mode=S_IFDIR|0755, st_size=4096, ...}) = 0
umask(0) = 022
umask(022) = 0
write(6, "\0", 1) = 1
stat64("./sbin", {st_mode=S_IFDIR|0755, st_size=4096, ...}) = 0
read(7, Segmentation fault
"", 1) = 0
--- SIGCHLD (Child exited) @ 0 (0) ---
read(7, "", 1) = 0
close(7) = 0
waitpid(10650, [{WIFEXITED(s) && WEXITSTATUS(s) == 139}], 0) = 10650
exit_group(1) = ?
Process 10649 detached
* Tobias Frost: Can you post a GDB stack trace? You should be able to obtain this by attaching gdb to the scp child process (with "gdb /usr/bin/scp PID") while ssh prompts for the password. Can you reproduce this with Debian's dropbear?
tags 448589 moreinfo thanks Or even telling strace to follow child processes (since it's the underlying ssh process that's segfaulting, not scp itself) as well as turning on scp's own debugging might help: strace -f scp -vvv -r root@fonera:/sbin . Thanks,
RESENT, forgot to cc to the BTS. sorry for the delay, the other mail got burried. Here are some traces, made with strace -ff -o scpsctrace scp -vvv -r root@fonera:/sbin . (scptrace.434xx) resp. strace -f -o scpsctrace scp -vvv -r root@fonera:/sbin . Hopefully this helps a little, I hope, I find some time soon to recompile it with debugging symbols to get a backtrace...