Bug#1037039: rsyslog - SysV init file missing
Package: rsyslog Version: 8.2302.0-1~bpo11+1 Can you please reinstate the missing /etc/init.d/rsyslog file in the package? >From the changelog notes, it appears to have been removed (intentionally) in version 8.2110.0-2. I'm not clear on why this is removed - one can continue to use a SysV init and so this file is useful/important in systems that do not use systemd to launch rsyslog. While I understand that ryslog's priority has been demoted from important to optional (per https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1018788), I see no reason to exclude the SysV init file from the package Thank you. Narayanan.
Bug#858084: closed by Ben Hutchings <b...@decadent.org.uk> (Re: Bug#858084: linux-image-amd64 - kernel option enable request)
On Thu, Mar 30, 2017 at 5:13 PM, Ben Hutchingswrote: > > > AFAIK, there's no (performance or size) penalty in enabling it [...] > > It's not very large, but yes there is a cost. > > Is it enough to leave it disabled? It seems that being a (standard) sysfs interface, the penalty is not really much at all (and considering all the other stuff that we do have enabled in the kernel, especially for convenience and/or usability, having this enabled shouldn't be an issue). I hope you can consider enabling it. Thanks. Regards, Narayanan.
Bug#858084: closed by Ben Hutchings <b...@decadent.org.uk> (Re: Bug#858084: linux-image-amd64 - kernel option enable request)
Hello, AFAIK, there's no (performance or size) penalty in enabling it and it provides some very convenient stats directly via sysfs. As a user of the stats, it is very convenient just to have them ready and usable without having to jump through hoops to access them. I don't understand the reason for not enabling them. Is there any reason *NOT* to enable it? Thanks. Regards, Narayanan. On Thu, Mar 30, 2017 at 4:45 AM, Debian Bug Tracking System < ow...@bugs.debian.org> wrote: > This is an automatic notification regarding your Bug report > which was filed against the linux-image-amd64 package: > > #858084: linux-image-amd64 - kernel option enable request > > It has been closed by Ben Hutchings <b...@decadent.org.uk>. > > Their explanation is attached below along with your original report. > If this explanation is unsatisfactory and you have not received a > better one in a separate message then please contact Ben Hutchings < > b...@decadent.org.uk> by > replying to this email. > > > -- > 858084: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=858084 > Debian Bug Tracking System > Contact ow...@bugs.debian.org with problems > > > -- Forwarded message -- > From: Ben Hutchings <b...@decadent.org.uk> > To: 858084-d...@bugs.debian.org > Cc: > Bcc: > Date: Thu, 30 Mar 2017 00:13:03 +0100 > Subject: Re: Bug#858084: linux-image-amd64 - kernel option enable request > On Sat, 2017-03-18 at 10:16 +0530, Narayanan R S wrote: > > Package: linux-image-amd64 > > Version: 4.9+79~bpo8+1 > > > > Can you please enable the CONFIG_CPU_FREQ_STAT_DETAILS option for the > > kernels. > > No, this is not really needed. You can use the power:cpu_frequency > tracepoint to see all transitions and then work out the statistics in a > user-space program (or use bpfcc to do it in-kernel). > > Ben. > > -- > Ben Hutchings > Everything should be made as simple as possible, but not simpler. >- Albert > Einstein > > > -- Forwarded message -- > From: Narayanan R S <n...@kadamba.org> > To: sub...@bugs.debian.org > Cc: > Bcc: > Date: Sat, 18 Mar 2017 10:16:20 +0530 > Subject: linux-image-amd64 - kernel option enable request > Package: linux-image-amd64 > Version: 4.9+79~bpo8+1 > > Can you please enable the CONFIG_CPU_FREQ_STAT_DETAILS option for the > kernels. > > Thanks in advance. > > Regards, > > Narayanan. > >
Bug#854677: linux-image-amd64 - patches from kernel source required
Hello, I think this bug can be closed as the kernel version has been updated which pretty much was the purpose of the original bug request. Thank you. Kind Regards, Narayanan.
Bug#858084: linux-image-amd64 - kernel option enable request
Package: linux-image-amd64 Version: 4.9+79~bpo8+1 Can you please enable the CONFIG_CPU_FREQ_STAT_DETAILS option for the kernels. Thanks in advance. Regards, Narayanan.
Bug#854677: linux-image-amd64 - patches from kernel source required
Hi Ben, I can try (re)running the 4.9 backport and see if the issue resurfaces. What is the correlation between the debian kernel (backported) and the mainline one (it says version 4.9.2-2~bpo8+1 - so is it essentially _EXACTLY_ 4.9.2)? Thanks. Regards, Narayanan. On Fri, Feb 17, 2017 at 8:50 AM, Ben Hutchings <b...@decadent.org.uk> wrote: > [Reply to all, not just to me.] > > On Fri, 2017-02-17 at 08:42 +0530, Narayanan R S wrote: > > Hello Ben, > > > > I did have the documented ICMP network issues when I used the 4.9 version > > of the backported kernel. > > > > When I reverted back to the 4.8 version (of the backported kernel), > > everything was back to normal. > > > > I am not sure when in the 4.9 series this was fixed but the bug report > > explicitly documents 4.9.5. I am not sure what version the backported > > kernel + patches corresponds to in the official/mainstream 4.9 kernel. > > The fix is said to have been applied *before* 4.9 and confirmed as > fixed in version 4.9.5. There have been no fixes to netfilter between > 4.9.2 (which I think you're using) and 4.9.5. > > So it sounds like you might have found a different bug. > > Ben. > > -- > Ben Hutchings > Any sufficiently advanced bug is indistinguishable from a feature. > >
Bug#854677: linux-image-amd64 - patches from kernel source required
Package: linux-image-amd64 Version: 4.9+78 The current (backported) version of linux-image-amd64 (for Jessie) appears to need a few important fixes from the kernel source (upstream). Please see: https://www.spinics.net/lists/kernel/msg2406212.html https://bugzilla.redhat.com/show_bug.cgi?id=1402695 It appears that this has been fixed in the 4.9 series after 4.9.5 (per the bugzilla link above) - so can the package be suitably updated to incorporate this fix? Thanks in advance. Regards, Narayanan.
Bug#703228: Any reasons for not including this fix?
Hi, Is there any reason for not including this fix in the current version of grub-legacy? I have tested this patch for both ext4 functionality and gpt functionality and it works perfectly. Even though these fixes are available in grub2, I think it is useful and doesn't require much effort to roll into grub-legacy. Can the maintainers please comment? If there's anything I can do to help please let me know. Thanks. Narayanan. -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#447556: wml: logos are completely broken - files are missing
Package: wml Version: 2.0.11-1 Severity: important wml::std::logo is totally broken as the logo files are missing. They should be present in /usr/lib/wml/data/logos/ but the logos directory is missing. Once you create the logos directory and put in the logo-*.info and logo-*.{gif|png} files things work fine. The original package is missing these files (and they are available from the source in the wml_misc directory). This probably took place between 2.0.8 and 2.0.11 when the logo locations was moved from .../aux/ to .../data/logos. This bug renders the entire wml::std::logo macro/package unusable. Narayanan. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#420933: ippl: patches to fix some critical bugs
Package: ippl Version: 1.4.14-8 Severity: important I've attached a new (additions on the existing diff file) diff file with some patches to fix a bunch of critical issues with ippl. The summary is: a) If you enable multiple protocol logs (eg: icmp+tcp+udp), there's a race condition due to which ippl will *NOT* run. The exact issue is that each thread drops (root) privileges after it attaches to the raw socket. If one of the other thread tries to attach after this drop in privileges it will fail and abort. The fix is to drop privileges in the main thread and not in each/every thread. There's a forced sleep in there to ensure that all the threads finish their socket attaches by then. b) As a result of (a) (and of course it was a bug from before), it is no longer possible to reload the config files and reattach to the socket since the privilege drop has already happened. Reload were broken and now I've explicitly logged a message to restart instead of reload (the process will quit gracefully after logging). The attached diff can be diffed with the previous diff from the source to get the exact changes I've made. Hopefully they are self explanatory. Please forward this upstream. Thanks. Narayanan. ippl_1.4.14-8.diff.gz Description: Diffs for ippl patches
Bug#420958: ipac-ng: patches to fix some critical bugs
Package: ipac-ng Version: 1.31-4 Severity: grave I've attached a patch (this is the modified diff from the original source distribution) to fix an issue with ipac-ng when trying to use accounting that involves ports(with tcp and/or udp). Currently ipac-ng is broken as it cannot deal with ports due to a library load issue. This cropped up as a result of the sarge to etch upgrade and appears to be related to the change in the iptables libraries. The summary of the change is the use of RTLD_LAZY instead of RTLD_NOW in the ipac-ng-1.31/agents/iptables/iptables.c file. There are two references and I've changed them both to use LAZY. I've actually modified the sf1193721.dpatch file in debian/patches/ as this is the file that really fixes a bunch of other issues as well. Without this fix, ipac-ng is completely broken and will _NOT_ even start if your config file has port matches for udp/tcp protocols. This patch fixes the issue and ipac-ng runs perfectly for me (as it used to on sarge). Please incorporate this (this exact fix can also be backported to the 1.27-5 branch in case 1.31-4 cannot be bumped into stable, which I think should happen!). The attached diff can be diffed with the previous diff from the source to get the exact changes I've made. Hopefully they are self explanatory. They are: ### $zdiff ipac-ng_1.31-4.diff.gz ipac-ng_1.31-4.diff.gz-ORIG 698c698 @@ -0,0 +1,4847 @@ --- @@ -0,0 +1,4845 @@ 947,948c947 +- if (dlopen(path, RTLD_NOW)) { ++ if (dlopen(path, RTLD_LAZY)) { --- + if (dlopen(path, RTLD_NOW)) { 1071,1072c1070 +- if (dlopen(path, RTLD_NOW)) { ++ if (dlopen(path, RTLD_LAZY)) { --- + if (dlopen(path, RTLD_NOW)) { ### Please forward this upstream. Thanks. Narayanan. ipac-ng_1.31-4.diff.gz Description: ipac-ng patch (diff modification)
Bug#418906: vpnc: keepalive not implemented correctly
Package: vpnc Version: 0.3.3+SVN20051028-3 Severity: important Architecture: i386 Debian Release: Stable (etch) vpnc doesn't correctly implement keepalives. The code ignores the user supplied keepalive time (seconds) and uses an internal poll timeout (which is 2 seconds). As a result, keep alives are sent every 2 seconds. I'm attaching a patch (against the _ORIGINAL_ tunip.c file), which also incorporates the debian specific patch to the same file. The fix is quick and easy and updates the keepalive variable with the current time after every keep alive packet is sent, thus refreshing the next keep alive cycle to the user supplied value. The exact two lines that need to change are: // Save the new timestamp keepalive = rekey = time(NULL); (this is in the vpnc_main_loop function in tunip.c). Please also forward this patch upstream to the vpnc codebase. Thanks. Narayanan. *** tunip.c 2007-04-12 01:53:47.0 -0400 --- tunip.c.orig2007-04-12 01:53:47.0 -0400 *** static int tun_send_ip(struct encap_meth *** 335,348 return 1; } ! static int encap_esp_new(struct encap_method *encap, int encap_fd) { #ifdef IP_HDRINCL int hincl = 1; #endif ! encap-fd = encap_fd; #ifdef IP_HDRINCL if (setsockopt(encap-fd, IPPROTO_IP, IP_HDRINCL, hincl, sizeof(hincl)) == -1) { --- 335,352 return 1; } ! static int encap_esp_new(struct encap_method *encap, unsigned char proto) { #ifdef IP_HDRINCL int hincl = 1; #endif ! encap-fd = socket(PF_INET, SOCK_RAW, proto); + if (encap-fd == -1) { + perror(socket(SOCK_RAW)); + return -1; + } #ifdef IP_HDRINCL if (setsockopt(encap-fd, IPPROTO_IP, IP_HDRINCL, hincl, sizeof(hincl)) == -1) { *** int encap_esp_recv_peer(struct encap_met *** 764,779 } static uint8_t volatile do_kill; ! #define VPNC_POLL_TIMEOUT 2000 ! ! static int vpnc_main_loop(struct peer_desc *peer, struct encap_method *meth, int tun_fd, const char *pidfile) { struct pollfd pollfds[2]; - time_t keepalive, rekey, now; - int rv = 1; - keepalive = rekey = time(NULL); pollfds[0].fd = tun_fd; pollfds[0].events = POLLIN; pollfds[1].fd = encap_get_fd(meth); --- 768,782 } static uint8_t volatile do_kill; + static uint8_t *kill_packet; + static size_t kill_packet_size; + static struct sockaddr *kill_dest; ! static void vpnc_main_loop(struct peer_desc *peer, struct encap_method *meth, int tun_fd, const char *pidfile) { + int sock; struct pollfd pollfds[2]; pollfds[0].fd = tun_fd; pollfds[0].events = POLLIN; pollfds[1].fd = encap_get_fd(meth); *** static int vpnc_main_loop(struct peer_de *** 783,810 int presult; do { ! presult = poll(pollfds, sizeof(pollfds) / sizeof(pollfds[0]), VPNC_POLL_TIMEOUT); } while (presult == -1 errno == EINTR !do_kill); if (presult == -1) { syslog(LOG_ERR, poll: %m); continue; } - now = time(NULL); - if (opt_nat_keepalive (now - keepalive) = opt_nat_keepalive) { - const char *payload = \xff; - sendto(meth-fd, payload, 1, 0, - (struct sockaddr *)peer-remote_sa-dest, - sizeof(peer-remote_sa-dest)); - // Save the new timestamp - keepalive = rekey = time(NULL); - } - - if (opt_rekey_interval now-rekey = opt_rekey_interval) { - rv = 0; - do_kill = 255; - } - if (pollfds[0].revents POLLIN) { int pack; --- 786,798 int presult; do { ! presult = poll(pollfds, sizeof(pollfds) / sizeof(pollfds[0]), -1); } while (presult == -1 errno == EINTR !do_kill); if (presult == -1) { syslog(LOG_ERR, poll: %m); continue; } if (pollfds[0].revents POLLIN) { int pack; *** static int vpnc_main_loop(struct peer_de *** 877,894 } } ! if (rv != 0) { ! tun_close(oursa-tun_fd, oursa-tun_name); ! if (pidfile) ! unlink(pidfile); /* ignore errors */ ! syslog(LOG_NOTICE, terminated); ! } ! return rv; } static void killit(int signum) { do_kill = signum; /* unused */ } static void write_pidfile(const char *pidfile) --- 865,886
Bug#418907: rdesktop: patch for segfaults with libx11-6 1.0.3-7 (stable)
Package: rdesktop Version: 1.5.0-1 Severity: grave Architecture: i386 Debian Release: Stable (etch) After a recent update on stable, rdesktop segfaults repeatedly and is completely unusable. The root cause of this issue is a change in libx11-6 (1.0.3-7) which had a recent security fix. The main issue is that the XCreateImage function has a change which impacts rdesktop (and other packages that depend on this library). I'm attaching a patch that fixes this issue (I've tested it and it works perfectly on etch and is very stable). The fix is against the xwin.c file (from the original rdesktop source distribution) in the rdesktop distribution. Please update this ASAP since this is a critical bug and renders rdesktop completely unusable unless you downgrade to libx11-6 1.0.3-6. Please forward the patch upstream. Thanks. Narayanan. *** xwin.c Wed Apr 11 12:31:43 2007 --- rdesktop-1.5.0/xwin.c Mon Aug 7 07:45:44 2006 *** ui_create_glyph(int width, int height, u *** 2413,2427 g_create_glyph_gc = XCreateGC(g_display, bitmap, 0, NULL); image = XCreateImage(g_display, g_visual, 1, ZPixmap, 0, (char *) data, !width, height, 8, 0); ! /* Patch to prevent Seg Faults - based on changes in libx11 !* to fix a security issue (CVE-2007-1667) !* See: http://lists.debian.org/debian-x/2007/04/msg00052.html !* See: http://www.nabble.com/Bug-418295:-vice-broken-by-libx11-security-update-t3544947.html !* !* ORIGINAL: width, height, 8, scanline); !* NEW : width, height, 8, 0); !*/ image-byte_order = MSBFirst; image-bitmap_bit_order = MSBFirst; XInitImage(image); --- 2413,2419 g_create_glyph_gc = XCreateGC(g_display, bitmap, 0, NULL); image = XCreateImage(g_display, g_visual, 1, ZPixmap, 0, (char *) data, !width, height, 8, scanline); image-byte_order = MSBFirst; image-bitmap_bit_order = MSBFirst; XInitImage(image); *** ui_desktop_restore(uint32 offset, int x, *** 3220,3254 { XImage *image; uint8 *data; - int bitmap_pad; offset *= g_bpp / 8; data = cache_get_desktop(offset, cx, cy, g_bpp / 8); if (data == NULL) return; - if (g_server_depth == 8) - { - bitmap_pad = 8; - } - else - { - bitmap_pad = g_bpp; - - if (g_bpp == 24) - bitmap_pad = 32; - } - image = XCreateImage(g_display, g_visual, g_depth, ZPixmap, 0, ! (char *) data, cx, cy, bitmap_pad, 0); ! /* Patch to prevent Seg Faults - based on changes in libx11 !* to fix a security issue (CVE-2007-1667) !* See: http://lists.debian.org/debian-x/2007/04/msg00052.html !* See: http://www.nabble.com/Bug-418295:-vice-broken-by-libx11-security-update-t3544947.html !* !* ORIGINAL: (char *) data, cx, cy, BitmapPad(g_display), cx * g_bpp / 8); !* NEW : (char *) data, cx, cy, BitmapPad(g_display), 0); !*/ if (g_ownbackstore) { --- 3212,3225 { XImage *image; uint8 *data; offset *= g_bpp / 8; data = cache_get_desktop(offset, cx, cy, g_bpp / 8); if (data == NULL) return; image = XCreateImage(g_display, g_visual, g_depth, ZPixmap, 0, !(char *) data, cx, cy, BitmapPad(g_display), cx * g_bpp / 8); if (g_ownbackstore) {
Bug#361867: ~/.raggle/feeds.yaml zero-sized after crash
On Tue, 11 Apr 2006, Michael Ablassmeier wrote: hi, and immediately after this crash, ~/.raggle/feeds.yaml was zero-sized. Luckily, there was still a usable ~/.raggle/feeds.yaml~ around. im sorry to say im unable to reproduce this bug. Do you have a custom configuration file you can send me? What did you do right before raggle crashed? Did you Mark items as Unread or did this bug hit you out of nowhere? Can you reproduce it with your backup feeds.yaml~? Hi - I've run into this bug a few times and I've managed to (symptomatically) infer that this happens (to me) ONLY when raggle is saving the feed configuration (automatically every now and then) AND I use a key press like the cursor keys to select another window (item/desc/feed) as the active window. It's happened often enough that I've avoided using the arrow keys as much as possible (I don't want to disable the feed save feature). I never got around to debugging it though. I should think that you should be able to reproduce it if you use the arrow keys to move around a bunch of feeds (unhighlighting them possibly - since you're reading them) around the time an auto save kicks in. HTH. Nars. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]