Segfault when extracting

[frithnanth]

Segfault when extracting

[frithnanth]

My bad: .extract() doesn't have to explicitly .free the entry structure, since it's already taken care of by the DESTROY submethod in Archive::Libarchive::Entry. I pushed the new version.

…

On Tue, Jan 10, 2017 at 5:58 PM, Jonathan Worthington < ***@***.***> wrote: I just got a segfault when using this library to extract files. It's a simple use, on a .tar.gz file: given await($tarball-location) { my Archive::Libarchive $archive .= new( operation => LibarchiveExtract, file => $_ ); $archive.extract; $archive.close; unlink $_; } Running it under perl6-valgrind-m I see that it seems to be due to mis-use of libarchive somehow; it crashes when calling archive_entry_free: ==4199== Invalid write of size 8 ==4199== at 0x22C10164: archive_wstring_free (in /usr/lib/x86_64-linux-gnu/libarchive.so.13.1.2) ==4199== by 0x22C13A4C: archive_mstring_clean (in /usr/lib/x86_64-linux-gnu/libarchive.so.13.1.2) ==4199== by 0x22BD4BD8: archive_entry_clear (in /usr/lib/x86_64-linux-gnu/libarchive.so.13.1.2) ==4199== by 0x22BD4C98: archive_entry_free (in /usr/lib/x86_64-linux-gnu/libarchive.so.13.1.2) ==4199== by 0x50670DD: ??? (in /home/jnthn/dev/MoarVM/install/lib/libmoar.so) ==4199== by 0x16125B8F: ??? ==4199== by 0x5067018: dc_callvm_call_x64 (in /home/jnthn/dev/MoarVM/install/lib/libmoar.so) ==4199== by 0x22BD4C8F: ??? (in /usr/lib/x86_64-linux-gnu/libarchive.so.13.1.2) ==4199== by 0x273148BF: ??? ==4199== Address 0xc56ac20 is 256 bytes inside a block of size 1,072 free'd ==4199== at 0x4C2EDEB: free (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so) ==4199== by 0x22BDE1CC: ??? (in /usr/lib/x86_64-linux-gnu/libarchive.so.13.1.2) ==4199== by 0x50670DD: ??? (in /home/jnthn/dev/MoarVM/install/lib/libmoar.so) ==4199== by 0x16125B8F: ??? ==4199== by 0x5067018: dc_callvm_call_x64 (in /home/jnthn/dev/MoarVM/install/lib/libmoar.so) ==4199== by 0x22C14A6F: ??? (in /usr/lib/x86_64-linux-gnu/libarchive.so.13.1.2) ==4199== by 0x1034EC4F: ??? ==4199== Block was alloc'd at ==4199== at 0x4C2FB55: calloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so) ==4199== by 0x22BD31D2: archive_entry_new2 (in /usr/lib/x86_64-linux-gnu/libarchive.so.13.1.2) ==4199== by 0x22BDD608: archive_read_new (in /usr/lib/x86_64-linux-gnu/libarchive.so.13.1.2) ==4199== by 0x50670DD: ??? (in /home/jnthn/dev/MoarVM/install/lib/libmoar.so) ==4199== by 0x16125B8F: ??? ==4199== by 0x5067018: dc_callvm_call_x64 (in /home/jnthn/dev/MoarVM/install/lib/libmoar.so) ==4199== by 0x22BDD5CF: ??? (in /usr/lib/x86_64-linux-gnu/libarchive.so.13.1.2) ==4199== by 0x2640D55F: ??? — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#3>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ADyA2NpW7wkyWI0dZqll0O6oo_njqdiFks5rQ7i7gaJpZM4Lfpgz> .

-- Fernando Santagata

[jnthn]

Thanks for the fast response. Unfortunately this doesn't seem to do it; I still get the SEGV and the output under Valgrind looks the same. (I changed the use statement in my code to have :ver<0.0.3> to make sure I really was testing the latest version.)

[frithnanth]

I tried this: my $tarball-location = start { 't/test.tar.gz' }; given await($tarball-location) { my Archive::Libarchive $archive .= new( operation => LibarchiveExtract, file => $_ ); $archive.extract; $archive.close; } and it extracted the files without crashing. Please send me a minimal test case that crashes, so I can look into it.

…

On Tue, Jan 10, 2017 at 6:28 PM, Jonathan Worthington < ***@***.***> wrote: Thanks for the fast response. Unfortunately this doesn't seem to do it; I still get the SEGV and the output under Valgrind looks the same. (I changed the use statement in my code to have :ver<0.0.3> to make sure I really was testing the latest version.) — You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub <#3 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ADyA2MS3FpggTvYwWedNF6LXu3RXak6rks5rQ7-_gaJpZM4Lfpgz> .

-- Fernando Santagata

[frithnanth]

I think I have found a probable cause for the segfault: I need to free the Entry object only if the operation is 'write', not on 'read' and 'extract'.
I released v.0.0.4, let me know if it fixes the Segfault.

[jnthn]

Yes, that fixes it; no segfault and it works correctly. Furthermore, valgrind is silent now, so looks like there are no errors. Many thanks! :-)