Bug#356352: libsysfs2 patch
On Sun, Jul 09, 2006 at 12:03:50AM +0200, Steinar H. Gunderson wrote: > Newest upstream is now 13.0. I've taken the patch, forward-ported it to 13.0 > and fixed a few errors (valgrind is very useful for this kind of work, by the > way). It works without crashing or giving valgrind hits (at least "hwinfo" > and "hwinfo --pci"), but it has a number of small memory leaks. (I suspect > the latter is not a consequence of the patch, though.) thanks! as stated before I was unable to tell if hwinfo gave me more or less informations than without the patch, though the patch looks rather easy I suspect it is working. Of course the best would be to hear from upstream but I doubt they care enough. filippo -- Filippo Giunchedi - http://esaurito.net PGP key: 0x6B79D401 random quote follows: Gretchen: Donnie Darko? What the hell kind of name is that? It's like some sort of superhero or something Donnie: What makes you think I'm not? -- from Donnie Darko (2001) signature.asc Description: Digital signature
Processed: Re: Bug#356352: libsysfs2 patch
Processing commands for [EMAIL PROTECTED]: > tags 356352 + patch Bug#356352: hwinfo - FTBFS: warning: implicit declaration of function 'sysfs_read_dir_links' Tags were: help Tags added: patch > thanks Stopping processing here. Please contact me if you need assistance. Debian bug tracking system administrator (administrator, Debian Bugs database) -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Bug#356352: libsysfs2 patch
tags 356352 + patch
thanks
On Mon, Jun 26, 2006 at 09:06:39PM +0200, Morten Werner Olsen wrote:
> The patch applies fine with 12.33 (newest upstream I found atm), but I
> get this error when running hwinfo (this is the shared libary one -
> similar source package that is in experimental, but with your patch and
> newer upstream release):
Newest upstream is now 13.0. I've taken the patch, forward-ported it to 13.0
and fixed a few errors (valgrind is very useful for this kind of work, by the
way). It works without crashing or giving valgrind hits (at least "hwinfo"
and "hwinfo --pci"), but it has a number of small memory leaks. (I suspect
the latter is not a consequence of the patch, though.)
/* Steinar */
--
Homepage: http://www.sesse.net/
Only in hwinfo-13.0.patched: .depend
Only in hwinfo-13.0.patched: hwinfo
Only in hwinfo-13.0.patched: hwinfo.o
Only in hwinfo-13.0.patched/src/hd: adb.o
Only in hwinfo-13.0.patched/src/hd: bios.o
Only in hwinfo-13.0.patched/src/hd: block.o
Only in hwinfo-13.0.patched/src/hd: braille.o
Only in hwinfo-13.0.patched/src/hd: cpu.o
Only in hwinfo-13.0.patched/src/hd: .depend
diff -ur hwinfo-13.0/src/hd/edd.c hwinfo-13.0.patched/src/hd/edd.c
--- hwinfo-13.0/src/hd/edd.c2006-06-29 16:03:37.0 +0200
+++ hwinfo-13.0.patched/src/hd/edd.c2006-07-08 23:47:31.0 +0200
@@ -53,9 +53,9 @@
bios_info_t *bt;
edd_info_t *ei;
- struct sysfs_directory *sf_dir;
- struct sysfs_directory *sf_dir_2;
- struct sysfs_link *sf_link;
+ struct dlist *sf_dir_list;
+ struct sysfs_attribute *node, *tmpattr;
+ char attrpath[SYSFS_PATH_MAX], linkpath[SYSFS_PATH_MAX];
for(u = 0; u < sizeof hd_data->edd / sizeof *hd_data->edd; u++) {
free_mem(hd_data->edd[u].sysfs_id);
@@ -63,69 +63,92 @@
memset(hd_data->edd, 0, sizeof hd_data->edd);
- sf_dir = sysfs_open_directory("/sys/firmware/edd");
-
- if(sf_dir) {
-if(!sysfs_read_all_subdirs(sf_dir)) {
- if(sf_dir->subdirs) {
-dlist_for_each_data(sf_dir->subdirs, sf_dir_2, struct sysfs_directory)
{
+ sf_dir_list = sysfs_open_directory_list("/sys/firmware/edd");
+
+ if(sf_dir_list) {
+dlist_for_each_data(sf_dir_list, node, struct sysfs_attribute) {
if(
-sscanf(sf_dir_2->name, "int13_dev%02x", &u) == 1 &&
+sscanf(node->name, "int13_dev%02x", &u) == 1 &&
u >= 0x80 &&
u <= 0xff
) {
+
edd_cnt++;
u -= 0x80;
ei = hd_data->edd + u;
-
-if(hd_attr_uint(sysfs_get_directory_attribute(sf_dir_2,
"sectors"), &ul0, 0)) {
+
+sprintf(attrpath, "/sys/firmware/edd/%s/sectors", node->name);
+tmpattr = sysfs_open_attribute(attrpath);
+if(hd_attr_uint(tmpattr, &ul0, 0)) {
ei->sectors = ul0;
}
+sysfs_close_attribute(tmpattr);
-if(hd_attr_uint(sysfs_get_directory_attribute(sf_dir_2,
"default_cylinders"), &ul0, 0)) {
+sprintf(attrpath, "/sys/firmware/edd/%s/default_cylinders",
node->name);
+tmpattr = sysfs_open_attribute(attrpath);
+if(hd_attr_uint(tmpattr, &ul0, 0)) {
ei->edd.cyls = ul0;
}
+sysfs_close_attribute(tmpattr);
-if(hd_attr_uint(sysfs_get_directory_attribute(sf_dir_2,
"default_heads"), &ul0, 0)) {
+sprintf(attrpath, "/sys/firmware/edd/%s/default_heads",
node->name);
+tmpattr = sysfs_open_attribute(attrpath);
+if(hd_attr_uint(tmpattr, &ul0, 0)) {
ei->edd.heads = ul0;
}
+sysfs_close_attribute(tmpattr);
-if(hd_attr_uint(sysfs_get_directory_attribute(sf_dir_2,
"default_sectors_per_track"), &ul0, 0)) {
+sprintf(attrpath,
"/sys/firmware/edd/%s/default_sectors_per_track", node->name);
+tmpattr = sysfs_open_attribute(attrpath);
+if(hd_attr_uint(tmpattr, &ul0, 0)) {
ei->edd.sectors = ul0;
}
+sysfs_close_attribute(tmpattr);
-if(hd_attr_uint(sysfs_get_directory_attribute(sf_dir_2,
"legacy_max_cylinder"), &ul0, 0)) {
+sprintf(attrpath, "/sys/firmware/edd/%s/legacy_max_cylinder",
node->name);
+tmpattr = sysfs_open_attribute(attrpath);
+if(hd_attr_uint(tmpattr, &ul0, 0)) {
ei->legacy.cyls = ul0 + 1;
}
+sysfs_close_attribute(tmpattr);
-if(hd_attr_uint(sysfs_get_directory_attribute(sf_dir_2,
"legacy_max_head"), &ul0, 0)) {
+sprintf(attrpath, "/sys/firmware/edd/%s/legacy_max_head",
node->name);
+tmpattr = sysfs_open_attribute(attrpath);
+if(hd_attr_uint(tmpattr, &ul0, 0)) {
ei->legacy.heads = ul0 + 1;
}
+sysfs_close_attribute(tmpattr);
-if(hd_attr_uint(sysfs_get_directory_attribute(sf_dir_2,
"legacy_sectors_per_track")
Bug#356352: libsysfs2 patch
On Mon, Jun 26, 2006 at 09:06:39PM +0200, Morten Werner Olsen wrote: > > The output from debian's hwinfo and 12.9 patched are not identical though, > > I can't tell if there's more information or less, I have it available for > > display. > > Thanks a lot for this patch! :) > > The patch applies fine with 12.33 (newest upstream I found atm), but I > get this error when running hwinfo (this is the shared libary one - > similar source package that is in experimental, but with your patch and > newer upstream release): > > [EMAIL PROTECTED]:~$ /usr/sbin/hwinfo > > usb.3.3: input*** glibc detected *** double free or corruption > > (fasttop): 0x0807dbf0 *** > Aborted indeed, the patch is nowhere near to be polished, I need to check it further :| It should be a start, though. I'm going to have a look at it also, but don't hold your breath :) filippo -- Filippo Giunchedi - http://esaurito.net PGP key: 0x6B79D401 random quote follows: Date: Tuesday, 2002/10/22 - 08:09 dselect proves the existence of Satan. It's the worst part of Debian. signature.asc Description: Digital signature
Bug#356352: libsysfs2 patch
On Tue, Jun 13, 2006 at 11:32:55PM +0200, Filippo Giunchedi wrote: > this: > http://people.debian.org/~filippo/patches/hwinfo-12.31_libsysfs2-0.1.patch > is my first attempt to port hwinfo to libsysfs2, the patch is against > 12.31 but applies to 12.9 with all hunks successful (just a few offsets). > Please test the patch if it works for you (at least if it finds the same > amount of hardware, or more :)) > The output from debian's hwinfo and 12.9 patched are not identical though, > I can't tell if there's more information or less, I have it available for > display. Thanks a lot for this patch! :) The patch applies fine with 12.33 (newest upstream I found atm), but I get this error when running hwinfo (this is the shared libary one - similar source package that is in experimental, but with your patch and newer upstream release): [EMAIL PROTECTED]:~$ /usr/sbin/hwinfo > usb.3.3: input*** glibc detected *** double free or corruption > (fasttop): 0x0807dbf0 *** Aborted This error is also when using --hw_item, when hw_item is one of block, bluetooth, braille, camera, cdrom, chipcard, disk, dsl, floppy, hub, isdn, joystick, keyboard, modem, mouse, netcard, network, partition, printer, scanner, usb and wlan. I also got a segmentation fault for hw_item like pci, pcmcia and pppoe: [EMAIL PROTECTED]:~$ /usr/sbin/hwinfo --pci > net.1: get network dataSegmentation fault I have not tested 12.33 without your patch yet, not as a static binary, and not 12.31 nor 12.9 with your patch. Hope to do some more testing later tonight, and will send updates to you and BTS. Maybe I will have time to run hwinfo in Valgrind too.. :) - Werner -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Bug#356352: libsysfs2 patch
Hi, this: http://people.debian.org/~filippo/patches/hwinfo-12.31_libsysfs2-0.1.patch is my first attempt to port hwinfo to libsysfs2, the patch is against 12.31 but applies to 12.9 with all hunks successful (just a few offsets). Please test the patch if it works for you (at least if it finds the same amount of hardware, or more :)) The output from debian's hwinfo and 12.9 patched are not identical though, I can't tell if there's more information or less, I have it available for display. thanks, filippo -- Filippo Giunchedi - http://esaurito.net PGP key: 0x6B79D401 random quote follows: What a strange illusion it is to suppose that beauty is goodness. -- Lev Tolstoj signature.asc Description: Digital signature
