Bug#1060099: telegram-desktop: FTBFS on mips64el: ./obj-mips64el-linux-gnuabi64/Telegram/./obj-mips64el-linux-gnuabi64/Telegram/gen/scheme.cpp:16435:(.text+0x1600a0): relocation truncated to fit: R_MI
Hello! On 06.01.2024 00:09:14 MSK you wrote: > CMakeFiles/td_scheme.dir/gen/scheme.cpp.o: in function `MTPDchannel::read(int > const*&, int const*)': > ./obj-mips64el-linux-gnuabi64/Telegram/./obj-mips64el-linux-gnuabi64/Telegram/gen/scheme.cpp:16435:(.text+0x1600a0): > relocation truncated to fit: R_MIPS_GOT_PAGE against `.text' > CMakeFiles/td_scheme.dir/gen/scheme.cpp.o: in function `MTPDuser::read(int > const*&, int const*)': > ./obj-mips64el-linux-gnuabi64/Telegram/./obj-mips64el-linux-gnuabi64/Telegram/gen/scheme.cpp:15827:(.text+0x16139c): > relocation truncated to fit: R_MIPS_GOT_PAGE against `.text' > collect2: error: ld returned 1 exit status Sorry, I have no idea how to properly fix this linker error. The aforementioned translation unit already uses compiler flags -mxgot -fPIC. They helped before. It seems the scheme.cpp has reached the limit of code size of single TU. Primary platform for Telegram Desktop is amd64; arm64 is also somewhat popular. This issue is platform specific. And I daresay it is not that serious to block release on unaffected architectures. signature.asc Description: This is a digitally signed message part.
Bug#1016744: Test failure has become significantly rarer
severity 1016744 minor retitle 1016744 vim-ultisnips: flaky autopkgtest with-neovim stop I applied a patch [1] increasing attempts of tests with regular expressions. And now probability of failure is far less. Although, the initial issue with a race of I/O in tmux(1) was not completely fixed. I daresay it does not block our release of the package. If you still encounter failure of the with-neovim autopkgtest as of package version 3.2-2, please retry the test and it should pass. Sorry for inconvenience. [1]: https://salsa.debian.org/mymedia/vim-ultisnips/-/commit/50df4908a8452042d358cce94527c22b4d253872 signature.asc Description: This is a digitally signed message part.
Bug#1018842: telegram-destop: FTBFS with libabsl20220623
Control: tag -1 ftbfs I haven't opportunity to check this yet. But it seems the libtgowt package needs to be rebuild as well. This package contains a static library and so brings the abseil dependency. For some reason, libtgowt is not in the transition plan, though there is libabsl-dev on the Depends list. signature.asc Description: This is a digitally signed message part.
Bug#1004815: qtox: FTBFS with ffmpeg 5.0
On March 7th, upstream author Anthony Bilinski released new version 1.17.6 addressing this issue. We should package it as far as possible in order to unblock FFmpeg transition. I am working on an NMU and discovered that Git packaging repository which d/control refers to, does not match source uploaded to Debian. However, both mention package version 1.17.4-1. signature.asc Description: This is a digitally signed message part.
Bug#1004637: libquicktime: FTBFS with ffmpeg 5.0
Control: tag -1 patch
Here is a possible patch. Not tested, though.
Upstream author decided to disable FFmpeg plugin with the new version of the
library. Perhaps, we have to do the same.
https://sourceforge.net/p/libquicktime/git/ci/974ad1489bb1d7d1ebc94f301508702f8ac2083b/
--- a/plugins/ffmpeg/audio.c
+++ b/plugins/ffmpeg/audio.c
@@ -423,8 +423,8 @@ static int a52_header_read(a52_header *
typedef struct
{
AVCodecContext * avctx;
- AVCodec * encoder;
- AVCodec * decoder;
+ AVCodec const * encoder;
+ AVCodec const * decoder;
int initialized;
@@ -512,7 +512,6 @@ static int decode_chunk_vbr(quicktime_t
#if DECODE_AUDIO4
AVFrame f;
- int got_frame;
#endif
chunk_packets = lqt_audio_num_vbr_packets(file, track, track_map->cur_chunk, &num_samples);
@@ -548,13 +547,14 @@ static int decode_chunk_vbr(quicktime_t
codec->pkt.size = packet_size + AV_INPUT_BUFFER_PADDING_SIZE;
#if DECODE_AUDIO4
-frame_bytes = avcodec_decode_audio4(codec->avctx, &f,
-&got_frame, &codec->pkt);
-if(frame_bytes < 0)
+if(avcodec_send_packet(codec->avctx, &codec->pkt) < 0 &&
+ avcodec_receive_frame(codec->avctx, &f) < 0)
{
lqt_log(file, LQT_LOG_ERROR, LOG_DOMAIN, "avcodec_decode_audio4 error");
break;
}
+frame_bytes = codec->pkt.size;
+
bytes_decoded = f.nb_samples * 2 * track_map->channels;
memcpy(&codec->sample_buffer[track_map->channels *
(codec->sample_buffer_end -
@@ -615,7 +615,6 @@ static int decode_chunk(quicktime_t * fi
#if DECODE_AUDIO4
AVFrame f;
- int got_frame;
#endif
/* Read chunk */
@@ -764,14 +763,14 @@ static int decode_chunk(quicktime_t * fi
codec->pkt.size = codec->bytes_in_chunk_buffer + AV_INPUT_BUFFER_PADDING_SIZE;
#if DECODE_AUDIO4
-
-frame_bytes = avcodec_decode_audio4(codec->avctx, &f,
-&got_frame, &codec->pkt);
-if(frame_bytes < 0)
+if(avcodec_send_packet(codec->avctx, &codec->pkt) < 0 ||
+ avcodec_receive_frame(codec->avctx, &f) < 0)
{
lqt_log(file, LQT_LOG_ERROR, LOG_DOMAIN, "avcodec_decode_audio4 error");
break;
}
+frame_bytes = codec->pkt.size;
+
bytes_decoded = f.nb_samples * 2 * track_map->channels;
memcpy(&codec->sample_buffer[track_map->channels *
(codec->sample_buffer_end -
@@ -1198,7 +1197,6 @@ static int lqt_ffmpeg_encode_audio(quick
#if ENCODE_AUDIO2
AVFrame f;
AVPacket pkt;
- int got_packet;
#endif
if(!codec->initialized)
@@ -1274,15 +1272,11 @@ static int lqt_ffmpeg_encode_audio(quick
codec->avctx->frame_size * channels * 2,
1);
-if(avcodec_encode_audio2(codec->avctx, &pkt,
- &f, &got_packet) < 0)
+if(avcodec_send_frame(codec->avctx, &f) < 0 ||
+ avcodec_receive_packet(codec->avctx, &pkt) < 0)
return 0;
-if(got_packet && pkt.size)
- frame_bytes = pkt.size;
-else
- frame_bytes = 0;
-
+frame_bytes = pkt.size;
#else
frame_bytes = avcodec_encode_audio(codec->avctx, codec->chunk_buffer,
codec->chunk_buffer_alloc,
@@ -1474,8 +1468,9 @@ static int read_packet_ac3(quicktime_t *
}
void quicktime_init_audio_codec_ffmpeg(quicktime_codec_t * codec_base,
- quicktime_audio_map_t *atrack, AVCodec *encoder,
- AVCodec *decoder)
+ quicktime_audio_map_t *atrack,
+ const AVCodec *encoder,
+ const AVCodec *decoder)
{
quicktime_ffmpeg_audio_codec_t *codec;
--- a/plugins/ffmpeg/ffmpeg.h
+++ b/plugins/ffmpeg/ffmpeg.h
@@ -30,10 +30,12 @@
void quicktime_init_video_codec_ffmpeg(quicktime_codec_t * codec,
quicktime_video_map_t *vtrack,
- AVCodec *encoder, AVCodec *decoder);
+ const AVCodec *encoder,
+ const AVCodec *decoder);
void quicktime_init_audio_codec_ffmpeg(quicktime_codec_t * codec,
quicktime_audio_map_t *vtrack,
- AVCodec *encoder, AVCodec *decoder);
+ const AVCodec *encoder,
+ const AVCodec *decoder);
void lqt_ffmpeg_set_parameter(AVCodecContext * ctx,
#if LIBAVCODEC_VERSION_MAJOR >= 54
--- a/plugins/ffmpeg/lqt_ffmpeg.c
+++ b/plugins/ffmpeg/lqt_ffmpeg.c
@@ -315,8 +315,8 @@ struct CODECIDMAP
{
int id;
int index;
- AVCodec *encoder;
- AVCodec *decoder;
+ AVCodec const *encoder;
+ AVCodec const *decoder;
lqt_parameter_info_static_t * encode_parameters;
lqt_parameter
Bug#1004821: libam7xxx: FTBFS with ffmpeg 5.0
This patch should help. Although, call to deprecated av_init_packet() remains.
It is to be replaced with av_packet_alloc() / av_packet_free() logic.--- a/examples/am7xxx-play.c
+++ b/examples/am7xxx-play.c
@@ -29,6 +29,7 @@
#include
#include
+#include
#include
#include
#include
@@ -49,17 +50,15 @@ static int video_input_init(struct video
const char *input_path,
AVDictionary **input_options)
{
- AVInputFormat *input_format = NULL;
+ AVInputFormat const *input_format = NULL;
AVFormatContext *input_format_ctx;
AVCodecParameters *input_codec_params;
AVCodecContext *input_codec_ctx;
- AVCodec *input_codec;
+ AVCodec const *input_codec;
int video_index;
int ret;
avdevice_register_all();
- avcodec_register_all();
- av_register_all();
if (input_format_string) {
/* find the desired input format */
@@ -158,7 +157,7 @@ static int video_output_init(struct vide
am7xxx_device *dev)
{
AVCodecContext *output_codec_ctx;
- AVCodec *output_codec;
+ AVCodec const *output_codec;
unsigned int new_output_width;
unsigned int new_output_height;
int ret;
signature.asc
Description: This is a digitally signed message part.
Bug#995659: coturn: FTBFS with OpenSSL 3.0
Hello!
On Sun, 3 Oct 2021 20:12:20 +0200 Kurt Roeckx wrote:
> src/client/ns_turn_msg.c: In function stun_produce_integrity_key_str:
> src/client/ns_turn_msg.c:260:7: warning: implicit declaration of function
> FIPS_mode [-Wimplicit-function-declaration]
> 260 | if (FIPS_mode()) {
> | ^
> [...]
> /usr/bin/ld: lib/libturnclient.a(ns_turn_msg.o): in function
> `stun_produce_integrity_key_str':
> ./src/client/ns_turn_msg.c:260: undefined reference to `FIPS_mode'
> collect2: error: ld returned 1 exit status
Please consider to apply my patch fixing build against newest OpenSSL.
The patch refers to the OPENSSL_FIPS macro that used to be in ancient
OpenSSL source code. This fix does not change behaviour because
the FIPS_mode() function always returned zero.
diffstat for coturn-4.5.2 coturn-4.5.2
changelog |8
patches/No-FIPS-140-mode.patch | 31 +++
patches/series |1 +
3 files changed, 40 insertions(+)
diff -Nru coturn-4.5.2/debian/changelog coturn-4.5.2/debian/changelog
--- coturn-4.5.2/debian/changelog 2021-03-24 22:30:23.0 +0300
+++ coturn-4.5.2/debian/changelog 2022-06-02 12:41:25.0 +0300
@@ -1,3 +1,11 @@
+coturn (4.5.2-3.1) unstable; urgency=medium
+
+ * Non-maintainer upload.
+ * New No-FIPS-140-mode.patch fixes build against OpenSSL 3.0
+ (Closes: #995659, LP: #1967018)
+
+ -- Nicholas Guriev Thu, 02 Jun 2022 12:41:25 +0300
+
coturn (4.5.2-3) unstable; urgency=medium
[ Ferenc Wágner ]
diff -Nru coturn-4.5.2/debian/patches/No-FIPS-140-mode.patch
coturn-4.5.2/debian/patches/No-FIPS-140-mode.patch
--- coturn-4.5.2/debian/patches/No-FIPS-140-mode.patch 1970-01-01
03:00:00.0 +0300
+++ coturn-4.5.2/debian/patches/No-FIPS-140-mode.patch 2022-06-02
12:34:17.0 +0300
@@ -0,0 +1,31 @@
+From: Nicholas Guriev
+Date: Thu, 02 Jun 2022 12:34:17 +0300
+Subject: Do not check FIPS 140 mode
+ It is not available in OpenSSL as packaged in Debian. The OPENSSL_FIPS macro
+ appeared in ancient OpenSSL sources but was never defined.
+ https://sources.debian.org/src/openssl/1.1.1n-0%2Bdeb11u2/crypto/o_fips.c/
+
+---
+ src/client/ns_turn_msg.c |4 ++--
+ 1 file changed, 2 insertions(+), 2 deletions(-)
+
+--- a/src/client/ns_turn_msg.c
b/src/client/ns_turn_msg.c
+@@ -244,7 +244,7 @@ int stun_produce_integrity_key_str(const
+ unsigned int keylen = 0;
+ EVP_MD_CTX ctx;
+ EVP_MD_CTX_init(&ctx);
+-#if defined EVP_MD_CTX_FLAG_NON_FIPS_ALLOW &&
!defined(LIBRESSL_VERSION_NUMBER)
++#ifdef OPENSSL_FIPS
+ if (FIPS_mode()) {
+
EVP_MD_CTX_set_flags(&ctx,EVP_MD_CTX_FLAG_NON_FIPS_ALLOW);
+ }
+@@ -256,7 +256,7 @@ int stun_produce_integrity_key_str(const
+ #else
+ unsigned int keylen = 0;
+ EVP_MD_CTX *ctx = EVP_MD_CTX_new();
+-#if defined EVP_MD_CTX_FLAG_NON_FIPS_ALLOW && !
defined(LIBRESSL_VERSION_NUMBER)
++#ifdef OPENSSL_FIPS
+ if (FIPS_mode()) {
+ EVP_MD_CTX_set_flags(ctx,
EVP_MD_CTX_FLAG_NON_FIPS_ALLOW);
+ }
diff -Nru coturn-4.5.2/debian/patches/series coturn-4.5.2/debian/patches/series
--- coturn-4.5.2/debian/patches/series 2021-02-10 22:05:50.0 +0300
+++ coturn-4.5.2/debian/patches/series 2022-06-02 12:34:17.0 +0300
@@ -1,2 +1,3 @@
Set-logging-to-syslog.patch
Fix-undefinded-OPENSSL_VERSION_1_1_1.patch
+No-FIPS-140-mode.patch
signature.asc
Description: This is a digitally signed message part
Bug#1005650: squid: FTBFS with OpenSSL 3.0
Control: tags -1 patch
Hello everyone interested!
To fix FTBFS of the squid package, I offer to apply my changes with the
-Wno-error=deprecated-declarations flag and an original patch. This will
allow us to proceed with the OpenSSL transition.
See also my MR on Salsa.
https://salsa.debian.org/squid-team/squid/-/merge_requests/20
diff -Nru squid-5.5/debian/changelog squid-5.5/debian/changelog
--- squid-5.5/debian/changelog 2022-04-15 15:39:54.0 +0300
+++ squid-5.5/debian/changelog 2022-05-31 23:13:38.0 +0300
@@ -1,3 +1,23 @@
+squid (5.5-1.1) unstable; urgency=medium
+
+ * Non-maintainer upload.
+
+ [ Nicholas Guriev ]
+ * Fixing build against OpenSSL 3.0 (Closes: #1005650, LP: #1946205)
+
+ * debian/rules
+- Do not fail on errors about deprecated declarations from OpenSSL.
+- Remove -Wall in CFLAGS from the debian/rules file since upstream build
+ scripts already pass this flag.
+
+ * debian/patches/
+- New 0006-Fix-build-against-OpenSSL-3-0.patch
+
+ [ Simon Deziel ]
+ * apparmor: allow reading /etc/ssl/openssl.cnf
+
+ -- Nicholas Guriev Tue, 31 May 2022 23:13:38 +0300
+
squid (5.5-1) unstable; urgency=medium
[ Amos Jeffries ]
diff -Nru squid-5.5/debian/patches/0006-Fix-build-against-OpenSSL-3-0.patch squid-5.5/debian/patches/0006-Fix-build-against-OpenSSL-3-0.patch
--- squid-5.5/debian/patches/0006-Fix-build-against-OpenSSL-3-0.patch 1970-01-01 03:00:00.0 +0300
+++ squid-5.5/debian/patches/0006-Fix-build-against-OpenSSL-3-0.patch 2022-05-31 22:31:08.0 +0300
@@ -0,0 +1,210 @@
+From: Nicholas Guriev
+Date: Tue, 31 May 2022 22:31:08 +0300
+Subject: Make build against OpenSSL-3.0 possible
+ In OpenSSL, the SSL_get_ex_new_index macro (substituted to
+ CRYPTO_get_ex_new_index) requires CRYPTO_EX_dup as the second callback. This
+ typedef, for some reason, has got an extra asterisk near void* within
+ arguments into the third version. Freely conversions from void* to void** is
+ okay in C but prohibited in C++. So I've updated the callback prototype to
+ match the last OpenSSL version.
+ .
+ OpenSSL pre-3.0 defined all of the SSL_OP_* macros with numeric hexadecimal
+ literals. However, the third version uses there casting expressions with
+ shifts which preprocessor is unable to compute. So I check only macros
+ existence, this lets Squid accept obsolete options. But it's nothing,
+ OpenSSL should ignore them anyway.
+
+---
+ acinclude/lib-checks.m4 |2 -
+ src/security/PeerOptions.cc | 50 ++--
+ src/ssl/support.cc |2 -
+ 3 files changed, 27 insertions(+), 27 deletions(-)
+
+--- a/acinclude/lib-checks.m4
b/acinclude/lib-checks.m4
+@@ -236,7 +236,7 @@ AC_DEFUN([SQUID_CHECK_OPENSSL_CONST_CRYP
+ AC_COMPILE_IFELSE([AC_LANG_PROGRAM([
+ #include
+
+-int const_dup_func(CRYPTO_EX_DATA *, const CRYPTO_EX_DATA *, void *, int, long, void *) {
++int const_dup_func(CRYPTO_EX_DATA *, const CRYPTO_EX_DATA *, void **, int, long, void *) {
+ return 0;
+ }
+ ],[
+--- a/src/security/PeerOptions.cc
b/src/security/PeerOptions.cc
+@@ -297,130 +297,130 @@ static struct ssl_option {
+
+ } ssl_options[] = {
+
+-#if SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG
++#ifdef SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG
+ {
+ "NETSCAPE_REUSE_CIPHER_CHANGE_BUG", SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG
+ },
+ #endif
+-#if SSL_OP_SSLREF2_REUSE_CERT_TYPE_BUG
++#ifdef SSL_OP_SSLREF2_REUSE_CERT_TYPE_BUG
+ {
+ "SSLREF2_REUSE_CERT_TYPE_BUG", SSL_OP_SSLREF2_REUSE_CERT_TYPE_BUG
+ },
+ #endif
+-#if SSL_OP_MICROSOFT_BIG_SSLV3_BUFFER
++#ifdef SSL_OP_MICROSOFT_BIG_SSLV3_BUFFER
+ {
+ "MICROSOFT_BIG_SSLV3_BUFFER", SSL_OP_MICROSOFT_BIG_SSLV3_BUFFER
+ },
+ #endif
+-#if SSL_OP_SSLEAY_080_CLIENT_DH_BUG
++#ifdef SSL_OP_SSLEAY_080_CLIENT_DH_BUG
+ {
+ "SSLEAY_080_CLIENT_DH_BUG", SSL_OP_SSLEAY_080_CLIENT_DH_BUG
+ },
+ #endif
+-#if SSL_OP_TLS_D5_BUG
++#ifdef SSL_OP_TLS_D5_BUG
+ {
+ "TLS_D5_BUG", SSL_OP_TLS_D5_BUG
+ },
+ #endif
+-#if SSL_OP_TLS_BLOCK_PADDING_BUG
++#ifdef SSL_OP_TLS_BLOCK_PADDING_BUG
+ {
+ "TLS_BLOCK_PADDING_BUG", SSL_OP_TLS_BLOCK_PADDING_BUG
+ },
+ #endif
+-#if SSL_OP_TLS_ROLLBACK_BUG
++#ifdef SSL_OP_TLS_ROLLBACK_BUG
+ {
+ "TLS_ROLLBACK_BUG", SSL_OP_TLS_ROLLBACK_BUG
+ },
+ #endif
+-#if SSL_OP_ALL
++#ifdef SSL_OP_ALL
+ {
+ "ALL", (long)SSL_OP_ALL
+ },
+ #endif
+-#if SSL_OP_SINGLE_DH_USE
++#ifdef SSL_OP_SINGLE_DH_USE
+ {
+ "SINGLE_DH_USE", SSL_OP_SINGLE_DH_USE
+ },
+ #endif
+-#if SSL_OP_EPHEMERAL_RSA
++#ifdef SSL_OP_EPHEMERAL_RSA
+ {
+ "EPHEMERAL_RSA", SSL_OP_EPHEMERAL_RSA
+ },
+ #endif
+-#if SSL_OP_PKCS1_CHECK_1
++#ifdef SSL_OP_PKCS1_CHECK_1
+ {
+ "PKCS1_CHECK_1", SSL_OP_PKCS1_CHECK_1
+ },
+ #endif
+-#if SSL_OP_PKC
Bug#1005718: mosh: FTBFS with OpenSSL 3.0
To prevent mosh from being deleted from testing in a few days, I am proposing to change warnings level to distcheck. This option effectively adds `-Wno-error=unused-parameter -Wno-error=deprecated-declarations` compiler flags. This solution will buy us time to rewrite related code before the deprecated low-level functions get removed away. diffstat for mosh-1.3.2 mosh-1.3.2 changelog |8 rules |2 +- 2 files changed, 9 insertions(+), 1 deletion(-) diff -Nru mosh-1.3.2/debian/changelog mosh-1.3.2/debian/changelog --- mosh-1.3.2/debian/changelog 2018-08-08 20:45:44.0 +0300 +++ mosh-1.3.2/debian/changelog 2022-05-28 10:01:51.0 +0300 @@ -1,3 +1,11 @@ +mosh (1.3.2-2.2) unstable; urgency=medium + + * Non-maintainer upload. + * Lower warning level to distcheck to solve build against OpenSSL 3.0 with +permitted deprecated declarations. (Closes: #1005718) + + -- Nicholas Guriev Sat, 28 May 2022 10:01:51 +0300 + mosh (1.3.2-2.1) unstable; urgency=low * Non-maintainer upload. diff -Nru mosh-1.3.2/debian/rules mosh-1.3.2/debian/rules --- mosh-1.3.2/debian/rules 2017-08-25 01:04:25.0 +0300 +++ mosh-1.3.2/debian/rules 2022-05-28 10:01:46.0 +0300 @@ -22,7 +22,7 @@ --disable-silent-rules \ --enable-ufw \ --enable-completion \ - --enable-compile-warnings=error + --enable-compile-warnings=distcheck override_dh_perl: # mosh only uses Perl modules in perl-base. signature.asc Description: This is a digitally signed message part
Bug#1011015: libkf5wayland-dev 4:5.93.0-1 breaks the include search of reverse dependencies
Control: clone -1 -2 Control: reassign -2 src:telegram-desktop 3.7.3+ds-1 Hello! On Sun, 15 May 2022 14:27:01 +0300 Adrian Bunk wrote: > https://buildd.debian.org/status/logs.php?pkg=telegram-desktop&ver=3.7.3%2Bds-1%2Bb1 > > ... > /<>/Telegram/lib_base/base/platform/linux/base_linux_wayland_integration.cpp:16:10: > fatal error: connection_thread.h: No such file or directory > 16 | #include > | ^ > compilation terminated. > make[3]: *** [Telegram/lib_base/CMakeFiles/lib_base.dir/build.make:224: > Telegram/lib_base/CMakeFiles/lib_base.dir/base/platform/linux/base_linux_wayland_integration.cpp.o] > Error 1 > > > > Both issues can be "fixed" by downgrading libkf5wayland-dev to 4:5.90.0-1 I do not believe this is a KWayland issue. Short paths were intentionally removed from there. https://invent.kde.org/frameworks/kwayland/-/commit/de442e4a94e249a29cf2e005db8e0a5e4a6a13ed As for Telegram Desktop I could provide a patch including the header. signature.asc Description: This is a digitally signed message part
Bug#1008156: telegram-desktop: Crash when switch account
Control: tags -1 - upstream On Пн, 2022-03-28 at 13:20 +0300, Nicholas Guriev wrote: > This will be fixed somehow in the upcoming upstream release, 3.6.2. I was wrong and did incorrect test. In actual fact, the crash happens due to memory allocator from glibc that is in use by default. Can you please check my workaround? Run the next commands in terminal: sudo apt-get install libjemalloc2 LD_PRELOAD=libjemalloc.so.2 telegram-desktop signature.asc Description: This is a digitally signed message part
Bug#1008156: telegram-desktop: Crash when switch account
Control: tag -1 upstream On Ср, 2022-03-23 at 11:49 +0100, Stefano Callegari wrote: > after last upgrade, when I try to switch to my second Telegram account (added > before), the program crashes. This will be fixed somehow in the upcoming upstream release, 3.6.2. signature.asc Description: This is a digitally signed message part
Bug#1005255: telegram-desktop: error: ‘Config’ is not a member of ‘webrtc’
Hello! The error in the tgcalls module happened due to an incompatibility with the libtgowt package. I was updating Telegram Desktop, and it is tied a much to the library, dedicated WebRTC fork. I uploaded a new upstream version yesterday and the error was gone. Sorry for the inconvenience. signature.asc Description: This is a digitally signed message part
Bug#1001977: src:telegram-desktop: fails to migrate to testing for too long: FTBFS on mipsel
Control: fixed -1 3.3.0+ds-2 Hello! The latest version I have uploaded is successfully built on every release architecture. And the package is now permitted to go to testing if no critical issues arise. signature.asc Description: This is a digitally signed message part
Bug#1001016: telegram-desktop backports request
Hello! On Пт, 2021-12-03 at 19:44 +, piorunz wrote: > Testing has got 3.1.1. I hope bullseye-backports can update it? CC to > maintainer, Nicholas Guriev! Thanks in advance. I have uploaded this version to Backports. It is already available on mirrors and should work okay. https://ftp.debian.org/debian/pool/main/t/telegram-desktop/ signature.asc Description: This is a digitally signed message part
Bug#1001016: telegram-desktop: Version in stable (2.6.1) becoming unusable
Hello! Could you test version 2.9.2 from backports? Does the message appear there? For a backports guide see: https://backports.debian.org/Instructions/ If even this version is considered ancient, I could backport 3.1.1 from testing. I am sure it still is working. From what I know, versions older that 3.1.1 are unable to operate for newly registered users whose UIDs do not fit signed 32bit. However, the older versions continue to work for already existing users. As a maintainer I can only offer you to switch to backports repository. Release managers do not happy with huge updates after Debian freeze. Telegram team in their turn is primarily focused on implementing new features. The goals seem contradictory, and users stuck between a rock and a hard place. signature.asc Description: This is a digitally signed message part
Bug#988885: marked as pending in rlottie
Control: tag -1 pending Hello, Bug #95 in rlottie reported by you has been fixed in the Git repository and is awaiting an upload. You can see the commit message below and you can check the diff of the fix at: https://salsa.debian.org/debian/rlottie/-/commit/2e80c28db4f1b0202b48c5afff1b394d4f7291f6 Sync patches with John Preston's fork * Also add Freetype-raster.patch for fix CVE-2021-31321. * Add Fortify-lottie-parser.patch for fix crashes on invalid input. Closes: #95 (this message was generated automatically) -- Greetings https://bugs.debian.org/95
Bug#988885: CVE-2021-31323 CVE-2021-31322 CVE-2021-31321 CVE-2021-31320 CVE-2021-31319 CVE-2021-31318 CVE-2021-31317 CVE-2021-31315
Hello! Thank you for pointing out these CVEs. I investigated deeper into the issues and reviewed the code as of 0.1+dfsg-1 version of the package. Luckily, most of these issues are not related to rlottie as currently packaged in Debian. Below are some of my notes. They do not imply 100% guarantee, and real tests are needed. CVE-2021-31323: Code was refactored. mData is now an std::vector this is extended before parseProperty() call. https://sources.debian.org/src/rlottie/0.1+dfsg-1/src/lottie/lottieparser.cpp/#L1741 CVE-2021-31322, CVE-2021-31319: Seems unaffected due to checking added by Fix-crash-on-invalid- data.patch https://sources.debian.org/src/rlottie/0.1+dfsg-1/src/lottie/lottiemodel.cpp/#L248 CVE-2021-31320: The mentioned while loop has been enhanced by Fix-crash-on-invalid- data.patch https://sources.debian.org/src/rlottie/0.1+dfsg-1/src/vector/vdrawhelper.cpp/#L168 CVE-2021-31318: Seems unaffected, because Fix-crash-on-invalid-data.patch inserts type checking before static_cast<> operator. https://sources.debian.org/src/rlottie/0.1+dfsg-1/src/lottie/lottieitem.cpp/#L454 CVE-2021-31315: Seems to be already fixed by Check-buffer-length.patch https://sources.debian.org/src/rlottie/0.1+dfsg-1/src/vector/vrle.cpp/#L559 CVE-2021-31321: Code differs, but bez_stack is an array of constant size on the gray_TWorker structure. It is twice the size of mentioned in the advisory. However, the vulnerability may be still present. https://sources.debian.org/src/rlottie/0.1+dfsg-1/src/vector/freetype/v_ft_raster.cpp/#L308 CVE-2021-31317: Not fixed. Need tests. As for the penultimate bug, I think it would be better to dispose of bundled freetype code and rely solely on libfreetype already packaged in Debian. But this may require a lot of changes that are unacceptable during freeze. Also note, these issues are all described in context of Telegram Android client. Nowadays, telegram-desktop is the only package in Debian main archive that depends on rlottie. Telegram Desktop does not support end- to-end encrypted secret chats, and so incoming animated stickers are subject to filtering by Telegram servers. Because of this, a remote attack is a little more difficult. There is another thing. For Debian, rlottie is built without a redefined RAPIDJSON_ASSERT macro, in contrast to upstream Telegram Desktop. By default the macro expands to abort() function call. This fact may result in additional SIGABRT crashes on invalid input data. But it will protect against more dangerous failures. https://github.com/desktop-app/cmake_helpers/blob/ac193a597d6b953f9869a240e21e275ce6e388cb/external/rlottie/CMakeLists.txt#L116 signature.asc Description: This is a digitally signed message part
Bug#981422: xdg-utils: autopkgtest failure: make: dh: No such file or directory
On Mon, 2021-03-01 at 10:40 +0100, Emilio Pozuelo Monfort wrote: > You can ship the files directly in the debian.tar by adding them to > debian/source/include-binaries, and that way you don't need to call chmod as > their mode will be preserved. That would probably help with this autopkgtest > patch issue. Unfortunately, dpkg does not see these new files because it first applies an autogenerated patch. At least, .debian.tar.xz archive has no files I mentioned in the include-binaries list. https://salsa.debian.org/freedesktop-team/xdg-utils/-/pipelines/235570 On Sun, 2021-02-28 at 10:20 -0300, Antonio Terceiro wrote: > Please drop patching from the test entirely. autopkgtests are supposed > to _start_ running against a fully patched source tree. autopkgtest(1) > does this automatically. > > I understand that this might be convenient when working on the package, > but please automate it in a way that it's not also done when running the > autopkgtest "for real". Any ideas how to preserve executable bit of new files in other way? For now, I see only one option: manually chmod +x either by the debian/rules patch target or repeated for dh_auto_test and in the autopkgtest runner. signature.asc Description: This is a digitally signed message part
Bug#981422: xdg-utils: autopkgtest failure: make: dh: No such file or directory
Hello!
On Sat, 2021-01-30 at 21:54 +0100, Paul Gevers wrote:
> You recently added an autopkgtest to your package xdg-utils, great.
> However, it fails. Currently this failure is blocking the migration to
> testing [1]. Can you please investigate the situation and fix it?
There are several ways to fix the failure.
1. Replace `debian/rules patch` line in a test script with `chmod +x
autotests/tty{on,off}`.
--- a/debian/tests/entry
+++ b/debian/tests/entry
@@ -13,6 +13,6 @@ fi
BASH_XTRACEFD=1
set -x
·
-debian/rules patch
./configure
+chmod +x autotests/tty{on,off}
make autotest SHELL="${1:-/bin/sh}"
https://salsa.debian.org/freedesktop-team/xdg-utils/-/commit/9928933504932f19fb39a0f671cdc7be78aada29
2. Put the "patch" target back as it was in -3 revision, and fix arch-
independent build.
https://salsa.debian.org/freedesktop-team/xdg-utils/-/merge_requests/4/diffs?commit_id=0617f9284ae4d79b51c4ad4bc7d781e93561cb23
I still prefer the later option with the "patch" target. Because that
chmod fixes flaws of quilt and dpkg and this solutions is more
universal. I hope someone will choose to upload another one-liner fix.
signature.asc
Description: This is a digitally signed message part
Bug#975228: bug marked as pending in xdg-utils
Control: tag -1 pending Hello, I made a fix of this bug by myself and committed it into the xdg-utils repository and now it is awaiting an upload. You can check the diff at: https://salsa.debian.org/freedesktop-team/xdg-utils/-/commit/628e0246c37260db118d71ffec47802f042b7773 signature.asc Description: This is a digitally signed message part
Bug#976037: marked as pending in ms-gsl
On Sat, 2020-12-26 at 15:59 +0100, Gianfranco Costamagna wrote: > if you want to test newer compilers, you are free to do it, but maybe you can > mark tests as flaky and > exit 77 if they fail, so at least you can see failures by manually looking at > logs... > > This way you won't block migration to testing, you won't be RC buggy, but you > will have the possibility to > have test logs. > > what do you think? In such case, I will not receive notifications about failed tests and so I can miss new errors. But I concluded that testing with GitLab CI on salsa.d.o would be better option. A test script for this can be written without need for release the package and it would not disturb anyone. On Sat, 2020-12-26 at 11:02 -0600, Steven Robbins wrote: > I believe there remain some corner cases where the build options of the > software under test may not precisely match those used in building googletest > -- leading to compile or test failures. One example is described in bug > #789267. > > That said, many projects do indeed successfully use the compiled library. So > you are probably fine. If you encounter odd failures, you can always revert > back to building gtest. Okay, thank you for comments. If something goes wrong, I will bear in mind possible differences in compilation flags. signature.asc Description: This is a digitally signed message part
Bug#976037: marked as pending in ms-gsl
I have rewritten auto-tests, so they do not longer require non-default versions of compilers. Since the tests with GCC rely on the same version of the compiler that built Google Test framework, I think preconditions of Bug#972944 lose its relevance, and there is no need to rebuild the googletest from sources (in /usr/src/googletest). Steve Robbins, am I right in my judgments? On Sat, 2020-12-26 at 13:43 +, Nicholas Guriev wrote: > Control: tag -1 pending > > Hello, > > Bug #976037 in ms-gsl reported by you has been fixed in the > Git repository and is awaiting an upload. You can see the commit > message below and you can check the diff of the fix at: > > https://salsa.debian.org/debian/ms-gsl/-/commit/66a9aa1dea09f4619e2ced31c8f277fda0eaa9b9 > > > Run autopkgtests only against default GCC and Clang > > * At Matthias Klose's request, removed auto-tests with fixed versions of GCC > or Clang. Instead, use default compilers. > * Rewritten the tests so that they include system-wide installed GSL headers. > > Closes: #976037 > > > (this message was generated automatically) signature.asc Description: This is a digitally signed message part
Bug#976037: marked as pending in ms-gsl
Control: tag -1 pending Hello, Bug #976037 in ms-gsl reported by you has been fixed in the Git repository and is awaiting an upload. You can see the commit message below and you can check the diff of the fix at: https://salsa.debian.org/debian/ms-gsl/-/commit/66a9aa1dea09f4619e2ced31c8f277fda0eaa9b9 Run autopkgtests only against default GCC and Clang * At Matthias Klose's request, removed auto-tests with fixed versions of GCC or Clang. Instead, use default compilers. * Rewritten the tests so that they include system-wide installed GSL headers. Closes: #976037 (this message was generated automatically) -- Greetings https://bugs.debian.org/976037
Bug#976982: telegram-desktop: Segmentation fault at start on Wayland GNOME
severity important merge 976982 976894 stop I daresay the bug you reported relates to screensaver integration that revealed today[1]. But if a workaround with QT_QPA_PLATFORM=wayland does not work for you, it could be something else. Can you please describe the crash more comprehensive and attach logs and terminal output of telegram-desktop to identify the issue? [1]: https://bugs.debian.org/976894 signature.asc Description: This is a digitally signed message part
Bug#976037: ms-gsl's autopkg tests are broken by design
Hello! On Sat, 28 Nov 2020 17:21:05 +0100 Matthias Klose wrote: > Now, there's at least one test (the clang-12 one), which never will succeed. So > what's the value of these tests when you always have failures due to > non-existing versions? Please just test for the default gcc and clang. The main idea is to let me discover issues as early as possible, before new toolchains become default. When a new compiler goes into archive, this CI test is triggered, although it does not block migrating when the compiler is not available yet due to "skip-not-installable" restriction. Testing against default GCC and Clang is already done on buildd trough the d/rules file and thanks to debhelper(7) which calls to `make test` during package build. Does it make sense to duplicate the work and do the same trough autopkgtest(1)?
Bug#966895: marked as pending in ms-gsl
Control: tag -1 pending Hello, Bug #966895 in ms-gsl reported by you has been fixed in the Git repository and is awaiting an upload. You can see the commit message below and you can check the diff of the fix at: https://salsa.debian.org/debian/ms-gsl/-/commit/c17b9395ae03e58493709e73b8ab794bebd8dcd7 New Move-array-bounds-tests.patch Closes: #966895 (this message was generated automatically) -- Greetings https://bugs.debian.org/966895
Bug#963320: libtgvoip: FTBFS: AttributeError: module 'string' has no attribute 'maketrans'
reassign 963320 gyp
found 963320 0.1+20200513gitcaa6002-1
affects 963320 src:libtgvoip
stop
On Sun, 2020-06-21 at 22:04 +0200, Lucas Nussbaum wrote:
> Hi,
>
> During a rebuild of all packages in sid, your package failed to build
> on amd64.
>
> Relevant part (hopefully):
> > make[1]: Entering directory '/<>'
> > gyp --format=cmake --depth=. --generator-output=. -Gconfig=Release
> > -Goutput_dir=obj-x86_64-linux-gnu
> > -Dlinux_path_opus_include=/usr/include/opus
> > Traceback (most recent call last):
> > File "/usr/bin/gyp", line 11, in
> > load_entry_point('gyp==0.1', 'console_scripts', 'gyp')()
> > File "/usr/lib/python3/dist-packages/gyp/__init__.py", line 552, in
> > script_main
> > return main(sys.argv[1:])
> > File "/usr/lib/python3/dist-packages/gyp/__init__.py", line 545, in main
> > return gyp_main(args)
> > File "/usr/lib/python3/dist-packages/gyp/__init__.py", line 518, in
> > gyp_main
> > [generator, flat_list, targets, data] = Load(
> > File "/usr/lib/python3/dist-packages/gyp/__init__.py", line 98, in Load
> > generator = __import__(generator_name, globals(), locals(),
> > generator_name)
> > File "/usr/lib/python3/dist-packages/gyp/generator/cmake.py", line 43, in
> >
> > _maketrans = string.maketrans
> > AttributeError: module 'string' has no attribute 'maketrans'
> > make[1]: *** [debian/rules:30: override_dh_auto_configure] Error 1
>
> The full build log is available from:
>http://qa-logs.debian.net/2020/06/20/libtgvoip_2.4.4-2_unstable.log
>
> A list of current common problems and possible solutions is available at
> http://wiki.debian.org/qa.debian.org/FTBFS . You're welcome to contribute!
>
> About the archive rebuild: The rebuild was done on EC2 VM instances from
> Amazon Web Services, using a clean, minimal and up-to-date chroot. Every
> failed build was retried once to eliminate random failures.
The bug is not in the libtgvoip package itself but apparently caused by
incomplete porting GYP to Python 3. Dear GYP maintainer, please fix
CMake generator, switch it to built-in str.maketrans.
I am already working on a new version of the libtgvoip package. It will
not use GYP and not solve the actual issue in the build tool.
Bug#959613: range-v3: FTBFS: unique.cpp:58:10: internal compiler error: in dependent_type_p, at cp/pt.c:25610
reassign 959613 g++-9 fixed 959613 9.3.0-10 found 959613 9.3.0-11 found 959613 9.3.0-13 severity 959613 important affects 959613 src:range-v3 stop Dear GCC maintainer, could you please take a little look at this compiler crash? The regression is appearing since GCC 9.3.0-11 uploaded on April 24 through 9.3.0-13, while with 9.3.0-10 the ragne-v3 package is built with no trouble.
Bug#959613: range-v3: FTBFS: unique.cpp:58:10: internal compiler error: in dependent_type_p, at cp/pt.c:25610
Eh... yet another GCC bug. I think the actual problem is not in the range-v3 package but in GCC. The compiler should be more robust and should not crash on any input files.
Bug#951302: timeshift: /mnt is used programmatically
Package: timeshift Version: 19.01+ds-2 Severity: serious Dear maintainer, Please modify the timeshift program so that it does not use /mnt/timeshift or any sub-directories under /mnt. FHS 3.0 says no package allowed to access this directory directly. It is intended to use as a temporally mount point by a system administrator[1]. Please switch to /run/timeshift instead. On first run, if a user has a Btrfs root file system on an encrypted device by LUKS and choose the BTRFS option in the Setup Wizard, the program automatically creates the /mnt/timeshift/backup directory and mount there the device with the root FS. On quit, this is not rolled back. Such behavior violates the Debian Policy and potentially may cause to data loss if there is something mounted by the user on their own. [1]: https://refspecs.linuxfoundation.org/FHS_3.0/fhs-3.0.html#mntMountPointForATemporarilyMount mymedia@barberry:~$ LANG=C sudo timeshift-gtk First run mode (config file not found) Selected default snapshot type: BTRFS App config loaded: /etc/timeshift.json Selected default snapshot device: /dev/sdb2 /dev/dm-2 is mounted at: /mnt/timeshift/backup, options: rw,relatime,ssd,space_cache,subvolid=5,subvol=/ /dev/dm-2 is mounted at: /mnt/timeshift/backup, options: rw,relatime,ssd,space_cache,subvolid=5,subvol=/ App config saved: /etc/timeshift.json App config saved: /etc/timeshift.json mymedia@barberry:~$ sudo rm /etc/timeshift.json -- System Information: Debian Release: buster/sid APT prefers eoan-updates APT policy: (500, 'eoan-updates'), (500, 'eoan-security'), (500, 'eoan') Architecture: amd64 (x86_64) Foreign Architectures: i386 Kernel: Linux 5.3.0-29-generic (SMP w/4 CPU cores) Kernel taint flags: TAINT_PROPRIETARY_MODULE, TAINT_OOT_MODULE, TAINT_UNSIGNED_MODULE Locale: LANG=ru_RU.UTF-8, LC_CTYPE=ru_RU.UTF-8 (charmap=UTF-8), LANGUAGE=ru (charmap=UTF-8) Shell: /bin/sh linked to /usr/bin/dash Init: systemd (via /run/systemd/system) LSM: AppArmor: enabled Versions of packages timeshift depends on: ii libc6 2.30-0ubuntu2 ii libcairo2 1.16.0-4 ii libgdk-pixbuf2.0-0 2.40.0+dfsg-1build1 ii libgee-0.8-20.20.2-1 ii libglib2.0-02.62.1-1 ii libgtk-3-0 3.24.12-1ubuntu1 ii libjson-glib-1.0-0 1.4.4-2 ii libvte-2.91-0 0.58.2-1ubuntu2 ii psmisc 23.2-1 ii rsync 3.1.3-6 timeshift recommends no packages. timeshift suggests no packages. -- no debconf information
Bug#929850: marked as pending in dhelp
Control: tag -1 pending Hello, Bug #929850 in dhelp reported by you has been fixed in the Git repository and is awaiting an upload. You can see the commit message below and you can check the diff of the fix at: https://salsa.debian.org/mymedia-guest/dhelp/commit/c2075fff759a2650b9272e5e37cec0daa514987c Do not remove entire /usr/share/doc/HTML Closes: #929850 (this message was generated automatically) -- Greetings https://bugs.debian.org/929850
Bug#926218: Bug #926218 in range-v3 marked as pending
Control: tag -1 pending Hello, Bug #926218 in range-v3 reported by you has been fixed in the Git repository and is awaiting an upload. You can see the commit message below and you can check the diff of the fix at: https://salsa.debian.org/debian/range-v3/commit/7c52e44a671922f23f960e5c6e0d79be893f771a Ignore false positive -Werror=maybe-uninitialized Closes: #926218 (this message was generated automatically) -- Greetings https://bugs.debian.org/926218
Bug#917315: Bug #917315 in libtgvoip marked as pending
Control: tag -1 pending Hello, Bug #917315 in libtgvoip reported by you has been fixed in the Git repository and is awaiting an upload. You can see the commit message below and you can check the diff of the fix at: https://salsa.debian.org/debian/libtgvoip/commit/989af959dd96fa167cc15d0dc5ecc8fe0a8d231e Fix porting issues Closes: #917315 (this message was generated automatically) -- Greetings https://bugs.debian.org/917315
Bug#917315: Bug #917315 in libtgvoip marked as pending
Control: tag -1 pending Hello, Bug #917315 in libtgvoip reported by you has been fixed in the Git repository and is awaiting an upload. You can see the commit message below and you can check the diff of the fix at: https://salsa.debian.org/debian/libtgvoip/commit/bc7b55d9f250ab40cca6dd4f32849fd6a5ba4a0e Fix porting issues Closes: #917315 (this message was generated automatically) -- Greetings https://bugs.debian.org/917315
Bug#906385: Bug #906385 in ms-gsl marked as pending
Control: tag -1 pending Hello, Bug #906385 in ms-gsl reported by you has been fixed in the Git repository and is awaiting an upload. You can see the commit message below, and you can check the diff of the fix at: https://salsa.debian.org/debian/ms-gsl/commit/b06f095c5b83f113e3cedd3beffde42e5c02f7c7 Add Catch-Classic-Workaround.patch Closes: #906385 (this message was generated automatically) -- Greetings https://bugs.debian.org/906385
Bug#904133: in telegram-destkop marked as pending
Control: tag -1 pending Hello, Bug #904133 in telegram-destkop reported by you has been fixed in the Git repository and is awaiting an upload. You can see the commit message below, and you can check the diff of the fix at: https://salsa.debian.org/debian/telegram-desktop/commit/53e93da1bb352eb25457afe9dc16336587cdacd7 Disable LTO at all * Unfortunately, this mode results to many FTBFS on 32-bit machines. * Nevertheless, on amd64 it reduces size of resulting binary by 22%. * Now less debug info is generated only for Release builds. Closes: #904133 (this message was generated automatically) -- Greetings https://bugs.debian.org/904133
Bug#889139: in telegram-desktop marked as pending
Control: tag 889139 pending Hello, Bug #889139 in telegram-desktop reported by you has been fixed in the Git repository. You can see the commit message below, and you can check the diff of the fix at: https://anonscm.debian.org/git/collab-maint/telegram-desktop.git/commit/?id=2c428e5 (this message was generated automatically based on the git commit message) --- commit 2c428e56a3419acec33316e39f7f391af9189bda Author: Adrian Bunk Date: Fri Feb 2 13:16:15 2018 +0200 Fix memory exhausted during linkiing on mips and mipsel Closes: #889139
Bug#859852: in telegram-desktop marked as pending
Control: tag 859852 pending Hello, Bug #859852 in telegram-desktop reported by you has been fixed in the Git repository. You can see the commit message below, and you can check the diff of the fix at: https://anonscm.debian.org/git/collab-maint/telegram-desktop.git/commit/?id=cacda19 (this message was generated automatically based on the git commit message) --- commit cacda19c0b3bfb84dcad3414fa713aef514e4636 Author: Nicholas Guriev Date: Sat Nov 11 21:45:35 2017 +0300 Add Use-OpenSSL-v1.1.patch Closes: #859852
Bug#871070: in ms-gsl marked as pending
Control: tag 871070 pending Hello, Bug #871070 in ms-gsl reported by you has been fixed in the Git repository. You can see the commit message below, and you can check the diff of the fix at: https://anonscm.debian.org/git/collab-maint/ms-gsl.git/commit/?id=d1efba7 (this message was generated automatically based on the git commit message) --- commit d1efba70a80f1fc771a1616a8b8262ddbbc491ff Author: Nicholas Guriev Date: Tue Aug 15 16:27:34 2017 +0300 Add Fix-tests-broken-with-GCC-7.patch Closes: #871070
