Bug #697
closed
broken libusb.so.1
0%
Description
in oi_148 library libusb.so.1 seems broken.
as an example, compiled apcupsd with usb support cannot be started (segmentation fault on launch) with usb in its config.
workaround: replace libusb.so.1 with file from oi_147
Files
Updated by Albert Lee over 12 years ago
- Category set to 10
- Priority changed from Low to Normal
- Target version changed from oi_148 to oi_151
Could be an existing bug exposed by building SFW with Studio 12.1.
debugging core file of gtkam (32-bit) from kirin file: /usr/bin/gtkam initial argv: gtkam threading model: native threads status: process terminated by SIGSEGV (Segmentation Fault), addr=f576361d > ::stack libc_hwcap1.so.1`strtok_r+0x54(f576361c, f5763c54, fee10380, fee9e334) libc_hwcap1.so.1`strtok+0x37(f576361c, f5763c54, 200, 0) libusb.so.1`usb_init+0x2d9(f6e70000, 50, fefc9374, f6e70268, feffb8f4, 8044cf4) usb.so`gp_port_library_list+0xda(8120e98) libgphoto2_port.so.0.8.0`foreach_func+0xcc(8242947, 8120e98) libltdl.so.3.1.4`foreachfile_callback+0x97(8241ba0, fdd22280, 8120e98) libltdl.so.3.1.4`foreach_dirinpath+0x294(fdd25da8, 0, f6b26910, fdd22280, 8120e98, 0) libltdl.so.3.1.4`lt_dlforeachfile+0x4c(fdd25da8, fdd22280, 8120e98, fdd2255e) libgphoto2_port.so.0.8.0`gp_port_info_list_load+0x88(8120e98, 81eb050, 8047178, 806be0e) gtkam_tree_load+0xb9(819a038, 80fea20, 80479b8, 805febd) main+0x2ec(1, 80479e8, 80479f0, 80479dc) _start+0x7d(1, 8047b08, 0, 8047b0e, 8047b1c, 8047b2f) > f576361c/s 0xf576361c: 1.1 > f5763c54/s 0xf5763c54: . > fee10380/x 0xfee10380: 4a72 > fee10380/X 0xfee10380: 8044a72 > 8044a72/s 0x8044a72: 1 >
Updated by Albert Lee about 12 years ago
The upstream bugs are 7002463 CUPS hp backend core in B153 and new queue cannot be added and 7007955.
Updated by Julian Wiesener about 12 years ago
- Assignee set to OI SFW
- Difficulty set to Medium
- Tags set to needs-triage
if i remember correctly, it is already fixed with 148b, needs to be checked
Updated by Gary Gendel about 12 years ago
This bug was fixed in oi_148b but is back with oi_151
Because of this, the apcusbd program crashes like it did in 147:
(gdb) where
#0 0xfee5c3ec in strtok_r () from /lib/libc.so.1
#1 0xfee5c37f in strtok () from /lib/libc.so.1
#2 0xfef42619 in usb_init () from /usr/lib/libusb.so.1
#3 0x08061077 in open_usb_device (ups=0x8081a58) at generic-usb.c:399
#4 0x080616da in pusb_ups_open (ups=0x8081a58) at generic-usb.c:632
#5 0x0805ef4a in usb_ups_open (ups=0x8081a58) at usb.c:220
#6 0x08055df3 in setup_device (ups=0x8081a58) at device.c:64
#7 0x080575d9 in main (argc=2, argv=0x8047c64) at apcupsd.c:285
I replaced /usr/lib/libusb.so.1 and /usr/lib/64/libusb.so.1 from 148b which solved the problem.
Updated by Alasdair Lumsden almost 12 years ago
- Priority changed from Normal to High
Updated by Alasdair Lumsden almost 12 years ago
Hi,
Is the fix in a mercurial patch queue?
Cheers,
Alasdair
Updated by Ken Mays almost 12 years ago
- File libusb_oi_151.tar.gz libusb_oi_151.tar.gz added
- Status changed from New to Feedback
Here are tested libusb 0.1.8 binaries tested for oi_151 as a workaround. Need inclusion in oi_151 respin as immediate patch or until valid resolution can be made.
Updated by Albert Lee almost 12 years ago
Someone needs to pull in the above mentioned fixes.
Updated by Andrzej Szeszo almost 12 years ago
Instead of using SFW built libusb and going through distro-import and distro-publish process I went ahead and re-used Oracle's userland build recipes to re-build libusb with Sun Studio 12. I have pushed updated package to /dev-il repo for testing.
gtkam seems to be happier with the updated library - it does not hang on startup
Build recipes can be found here for now (temporary location):
http://hg.openindiana.org/aszeszo/oi-extras/file/6eb7b62e444c/oi-extras/libusb/wrapper
Updated by Albert Lee almost 12 years ago
- Status changed from Feedback to Resolved
Fixed (reopen if we do rebuild with a newer compiler and don't have the proper fix integrated).