Re: KZoneAllocator static in KCompletion crashes on exit
On Monday 10 February 2014 01:28:03 Thomas Lübking wrote: On Sonntag, 9. Februar 2014 22:16:16 CEST, Albert Astals Cid wrote: So while exiting krecipes i get ==15297== Invalid read of size 8 ==15297==at 0xABE051C: QString::vsprintf(char const*, __va_list_tag*) (in /usr/lib/x86_64-linux-gnu/libQtCore.so.4.8.6) ==15297==by 0xAB87F94: qt_message(QtMsgType, char const*, __va_list_tag*) (in /usr/lib/x86_64-linux-gnu/libQtCore.so.4.8.6) ==15297==by 0xAB88190: qDebug(char const*, ...) (in /usr/lib/x86_64-linux-gnu/libQtCore.so.4.8.6) ==15297==by 0xA59AC14: KZoneAllocator::~KZoneAllocator() (kallocator.cpp:110) https://git.reviewboard.kde.org/r/114715/ ? Jep, that one removes the qDebug, which at that point must not be used anymore. So fixed already? Kevin, can you double-check? Cheers -- Milian Wolff m...@milianw.de http://milianw.de
Re: KZoneAllocator static in KCompletion crashes on exit
Am Montag, 10. Februar 2014, 01:28:03 schrieb Thomas Lübking: On Sonntag, 9. Februar 2014 22:16:16 CEST, Albert Astals Cid wrote: So while exiting krecipes i get ==15297== Invalid read of size 8 ==15297==at 0xABE051C: QString::vsprintf(char const*, __va_list_tag*) (in /usr/lib/x86_64-linux-gnu/libQtCore.so.4.8.6) ==15297==by 0xAB87F94: qt_message(QtMsgType, char const*, __va_list_tag*) (in /usr/lib/x86_64-linux-gnu/libQtCore.so.4.8.6) ==15297==by 0xAB88190: qDebug(char const*, ...) (in /usr/lib/x86_64-linux-gnu/libQtCore.so.4.8.6) ==15297==by 0xA59AC14: KZoneAllocator::~KZoneAllocator() (kallocator.cpp:110) https://git.reviewboard.kde.org/r/114715/ ? Cheers, Thomas Hey Albert, (Actually I replied much earlier, but KMail refused to send the mail somehow) It's interesting that you can reproduce this on a Linux system, I never got that far. What version of kdelibs are you running? This particular issue should indeed be fixed in v4.12.2 and KF5. The KZoneAllocator instance in KCompTreeNode got turned into 'static QSharedPointerKZoneAllocator alloc' member and KCompTreeNode instances are now keeping strong-refs to it to avoid the allocator (along its provided memory pool) being deleted too early. I also replaced the qDebug usage with printfs because I got similar crashes at shutdown. Greets -- Kevin Funk
Re: Review Request 115497: Replace SHA with PBKDF2-SHA512+Salt
--- This is an automatically generated e-mail. To reply, visit: https://git.reviewboard.kde.org/r/115497/ --- (Updated Feb. 10, 2014, 5:43 p.m.) Review request for KDE Runtime, Teo Mrnjavac and Valentin Rusu. Repository: kde-runtime Description --- Uses the MINOR_VERSION (which until now it was 0) to upgrade the hash from SHA to PBKDF2-SHA512+salt. I would have loved to completely replace it once the wallet is ported to the new hashing but because of kwalletd code that is not possible without a bigger rewrite. There are 2 reasons for this patch: 1-We avoid using our own implementation of SHA 2-We use a modern hashing technique I'm cooking more patches to use the system user password to open the wallet, we want that password to be hashed using PBKDF2_SHA512 for security reasons. Diffs (updated) - CMakeLists.txt 275a6c7 cmake/modules/FindLibGcrypt.cmake PRE-CREATION kwalletd/backend/CMakeLists.txt 5a5837c kwalletd/backend/backendpersisthandler.cpp bdef6ca kwalletd/backend/kwalletbackend.h 83ebf7f kwalletd/backend/kwalletbackend.cc e4d461c Diff: https://git.reviewboard.kde.org/r/115497/diff/ Testing --- Thanks, Àlex Fiestas
Re: KDE Review - Moving Artikulate to KDE Edu
El Diumenge, 9 de febrer de 2014, a les 17:42:55, Andreas Cord-Landwehr va escriure: On Sunday 26 January 2014 16:39:08 Albert Astals Cid wrote: El Dimarts, 14 de gener de 2014, a les 19:52:57, Albert Astals Cid va So I did have a look at Artikulate yesterday and just found some minor nagging user flow issues that I told Andreas and he says he would be fixing. Yeah, all fixed now. Other than that, looks good to me to move to kdeedu *but* we need to decide what to do with kqmlgraphplugin first, since we can't have a kdeedu app depending on a playground lib. Since kqtquickcharts is in KDE Review since the 5th, if no big issues are found it should migrate at the 19th to KDE Edu. So this problem will be solved 7 days before the freeze :) +1 Cheers, Albert Greetings, Andreas
Re: Review Request 114715: Attempt to fix KZoneAllocator issue
--- This is an automatically generated e-mail. To reply, visit: https://git.reviewboard.kde.org/r/114715/#review49478 --- This review has been submitted with commit c34045e2a2263865c825927de44c51faf5926132 by Albert Astals Cid on behalf of Kevin Funk to branch KDE/4.12. - Commit Hook On Jan. 28, 2014, 1:22 a.m., Kevin Funk wrote: --- This is an automatically generated e-mail. To reply, visit: https://git.reviewboard.kde.org/r/114715/ --- (Updated Jan. 28, 2014, 1:22 a.m.) Review request for kdelibs and Frank Reininghaus. Bugs: 327599 http://bugs.kde.org/show_bug.cgi?id=327599 Repository: kdelibs Description --- Attempt to fix KZoneAllocator issue kcompletion.p_h: Make the static KZoneAllocator member of KCompTreeNode a shared pointer so external users can hold a reference to it. kcompletion.cpp: Hold a reference to KCompTreeNode's KZoneAllocator instance so we avoid deleting the KZoneAllocator too early. See attached bug report for possible causes. (Hint: It crashes on Windows because ~KZoneAllocator is called to early.) kallocator.cpp: Use printf instead of qDebug(), because this code path code might be called very late during destruction and qDebug() will crash deep inside Qt. Also see discussion: http://lists.kde.org/?l=kde-develm=138583383708455w=1 BUG: 327599 CCBUG: 243375 Diffs - kdecore/util/kallocator.cpp 8b21120c62c513ea41686fe8185ec2808fe5d83a kdeui/util/kcompletion.cpp 340aa92b900d670e2ad73f70a63d5221d0feed1d kdeui/util/kcompletion_p.h 1cf31db3f16fe3421415cd54265eee20bb998710 Diff: https://git.reviewboard.kde.org/r/114715/diff/ Testing --- Thanks, Kevin Funk
Re: KZoneAllocator static in KCompletion crashes on exit
El Dilluns, 10 de febrer de 2014, a les 12:14:31, Kevin Funk va escriure: Am Montag, 10. Februar 2014, 01:28:03 schrieb Thomas Lübking: On Sonntag, 9. Februar 2014 22:16:16 CEST, Albert Astals Cid wrote: So while exiting krecipes i get ==15297== Invalid read of size 8 ==15297==at 0xABE051C: QString::vsprintf(char const*, __va_list_tag*) (in /usr/lib/x86_64-linux-gnu/libQtCore.so.4.8.6) ==15297==by 0xAB87F94: qt_message(QtMsgType, char const*, __va_list_tag*) (in /usr/lib/x86_64-linux-gnu/libQtCore.so.4.8.6) ==15297==by 0xAB88190: qDebug(char const*, ...) (in /usr/lib/x86_64-linux-gnu/libQtCore.so.4.8.6) ==15297==by 0xA59AC14: KZoneAllocator::~KZoneAllocator() (kallocator.cpp:110) https://git.reviewboard.kde.org/r/114715/ ? Cheers, Thomas Hey Albert, (Actually I replied much earlier, but KMail refused to send the mail somehow) It's interesting that you can reproduce this on a Linux system, I never got that far. What version of kdelibs are you running? This particular issue should indeed be fixed in v4.12.2 and KF5. Actually it was only fixed for master (4.13) and KF5. I've now cherry-picked it to KDE/4.12 too since it fixes the crash i was getting. Thanks! Cheers, Albert The KZoneAllocator instance in KCompTreeNode got turned into 'static QSharedPointerKZoneAllocator alloc' member and KCompTreeNode instances are now keeping strong-refs to it to avoid the allocator (along its provided memory pool) being deleted too early. I also replaced the qDebug usage with printfs because I got similar crashes at shutdown. Greets
Re: Review Request 115492: Add WebP detection from content in the mime package file
On Feb. 8, 2014, 5:41 p.m., David Faure wrote: Please consider submitting this mimetype to freedesktop.org.xml instead. https://bugs.freedesktop.org David Faure wrote: OK, s/instead/in addition/. Alex Merry wrote: Sorry, failed to record a conversation that happened on irc. Basically: https://bugs.freedesktop.org/show_bug.cgi?id=41155 Given the outcome of the bug, should we go ahead and implement our own thing? - Àlex --- This is an automatically generated e-mail. To reply, visit: https://git.reviewboard.kde.org/r/115492/#review49268 --- On Feb. 5, 2014, 2:44 p.m., Jerome Leclanche wrote: --- This is an automatically generated e-mail. To reply, visit: https://git.reviewboard.kde.org/r/115492/ --- (Updated Feb. 5, 2014, 2:44 p.m.) Review request for KDE Runtime and Alex Merry. Repository: kde-runtime Description --- This adds match detection for webp files (both lossy and lossless), the same way as audio/x-wav files are detected (both are RIFF based) Diffs - kimgio/webp.xml 4246f76 Diff: https://git.reviewboard.kde.org/r/115492/diff/ Testing --- Thanks, Jerome Leclanche
Re: Review Request 115497: Replace SHA with PBKDF2-SHA512+Salt
--- This is an automatically generated e-mail. To reply, visit: https://git.reviewboard.kde.org/r/115497/#review49485 --- - Valentin Rusu On Feb. 10, 2014, 5:43 p.m., Àlex Fiestas wrote: --- This is an automatically generated e-mail. To reply, visit: https://git.reviewboard.kde.org/r/115497/ --- (Updated Feb. 10, 2014, 5:43 p.m.) Review request for KDE Runtime, Teo Mrnjavac and Valentin Rusu. Repository: kde-runtime Description --- Uses the MINOR_VERSION (which until now it was 0) to upgrade the hash from SHA to PBKDF2-SHA512+salt. I would have loved to completely replace it once the wallet is ported to the new hashing but because of kwalletd code that is not possible without a bigger rewrite. There are 2 reasons for this patch: 1-We avoid using our own implementation of SHA 2-We use a modern hashing technique I'm cooking more patches to use the system user password to open the wallet, we want that password to be hashed using PBKDF2_SHA512 for security reasons. Diffs - CMakeLists.txt 275a6c7 cmake/modules/FindLibGcrypt.cmake PRE-CREATION kwalletd/backend/CMakeLists.txt 5a5837c kwalletd/backend/backendpersisthandler.cpp bdef6ca kwalletd/backend/kwalletbackend.h 83ebf7f kwalletd/backend/kwalletbackend.cc e4d461c Diff: https://git.reviewboard.kde.org/r/115497/diff/ Testing --- Thanks, Àlex Fiestas
Re: Review Request 115497: Replace SHA with PBKDF2-SHA512+Salt
--- This is an automatically generated e-mail. To reply, visit: https://git.reviewboard.kde.org/r/115497/#review49487 --- kwalletd/backend/kwalletbackend.cc https://git.reviewboard.kde.org/r/115497/#comment34903 The error logic should be more strict here, I think. In fact we loose the unknown version return code. - Valentin Rusu On Feb. 10, 2014, 5:43 p.m., Àlex Fiestas wrote: --- This is an automatically generated e-mail. To reply, visit: https://git.reviewboard.kde.org/r/115497/ --- (Updated Feb. 10, 2014, 5:43 p.m.) Review request for KDE Runtime, Teo Mrnjavac and Valentin Rusu. Repository: kde-runtime Description --- Uses the MINOR_VERSION (which until now it was 0) to upgrade the hash from SHA to PBKDF2-SHA512+salt. I would have loved to completely replace it once the wallet is ported to the new hashing but because of kwalletd code that is not possible without a bigger rewrite. There are 2 reasons for this patch: 1-We avoid using our own implementation of SHA 2-We use a modern hashing technique I'm cooking more patches to use the system user password to open the wallet, we want that password to be hashed using PBKDF2_SHA512 for security reasons. Diffs - CMakeLists.txt 275a6c7 cmake/modules/FindLibGcrypt.cmake PRE-CREATION kwalletd/backend/CMakeLists.txt 5a5837c kwalletd/backend/backendpersisthandler.cpp bdef6ca kwalletd/backend/kwalletbackend.h 83ebf7f kwalletd/backend/kwalletbackend.cc e4d461c Diff: https://git.reviewboard.kde.org/r/115497/diff/ Testing --- Thanks, Àlex Fiestas
Re: Review Request 115497: Replace SHA with PBKDF2-SHA512+Salt
--- This is an automatically generated e-mail. To reply, visit: https://git.reviewboard.kde.org/r/115497/#review49490 --- Ship it! One the issue around minor version check is done, feel free to commit it! - Valentin Rusu On Feb. 10, 2014, 5:43 p.m., Àlex Fiestas wrote: --- This is an automatically generated e-mail. To reply, visit: https://git.reviewboard.kde.org/r/115497/ --- (Updated Feb. 10, 2014, 5:43 p.m.) Review request for KDE Runtime, Teo Mrnjavac and Valentin Rusu. Repository: kde-runtime Description --- Uses the MINOR_VERSION (which until now it was 0) to upgrade the hash from SHA to PBKDF2-SHA512+salt. I would have loved to completely replace it once the wallet is ported to the new hashing but because of kwalletd code that is not possible without a bigger rewrite. There are 2 reasons for this patch: 1-We avoid using our own implementation of SHA 2-We use a modern hashing technique I'm cooking more patches to use the system user password to open the wallet, we want that password to be hashed using PBKDF2_SHA512 for security reasons. Diffs - CMakeLists.txt 275a6c7 cmake/modules/FindLibGcrypt.cmake PRE-CREATION kwalletd/backend/CMakeLists.txt 5a5837c kwalletd/backend/backendpersisthandler.cpp bdef6ca kwalletd/backend/kwalletbackend.h 83ebf7f kwalletd/backend/kwalletbackend.cc e4d461c Diff: https://git.reviewboard.kde.org/r/115497/diff/ Testing --- Thanks, Àlex Fiestas
Re: Review Request 115250: Try PASV mode when using Socks proxy
--- This is an automatically generated e-mail. To reply, visit: https://git.reviewboard.kde.org/r/115250/#review49496 --- Ship it! Ship It! - Dawit Alemayehu On Jan. 23, 2014, 11:31 a.m., Emil Sedgh wrote: --- This is an automatically generated e-mail. To reply, visit: https://git.reviewboard.kde.org/r/115250/ --- (Updated Jan. 23, 2014, 11:31 a.m.) Review request for kdelibs, Dawit Alemayehu and David Faure. Repository: kdelibs Description --- FTP has two modes: PASV and EPSV. Not all server's support EPSV. Currently, kio_ftp gives up on PASV mode if socks proxy is enabled. That is because QHostAddress.protocol() returns -1 (unknown protocol) on KUrl(socks://localhost:3128). So kio_ftp fails using SOCKS proxy on server's that lack EPSV support. This patch makes sure kio_ftp will try PASV mode if socks proxy is enabled. Diffs - kioslave/ftp/ftp.cpp 5bb2e8d Diff: https://git.reviewboard.kde.org/r/115250/diff/ Testing --- Tested a server that lacks EPSV support with and without proxy. Seems fine now. Used to throw 'Internal Server error'. Thanks, Emil Sedgh
Re: KZoneAllocator static in KCompletion crashes on exit
Am Montag, 10. Februar 2014, 20:38:15 schrieb Albert Astals Cid: El Dilluns, 10 de febrer de 2014, a les 12:14:31, Kevin Funk va escriure: Am Montag, 10. Februar 2014, 01:28:03 schrieb Thomas Lübking: On Sonntag, 9. Februar 2014 22:16:16 CEST, Albert Astals Cid wrote: So while exiting krecipes i get ==15297== Invalid read of size 8 ==15297==at 0xABE051C: QString::vsprintf(char const*, __va_list_tag*) (in /usr/lib/x86_64-linux-gnu/libQtCore.so.4.8.6) ==15297==by 0xAB87F94: qt_message(QtMsgType, char const*, __va_list_tag*) (in /usr/lib/x86_64-linux-gnu/libQtCore.so.4.8.6) ==15297==by 0xAB88190: qDebug(char const*, ...) (in /usr/lib/x86_64-linux-gnu/libQtCore.so.4.8.6) ==15297==by 0xA59AC14: KZoneAllocator::~KZoneAllocator() (kallocator.cpp:110) https://git.reviewboard.kde.org/r/114715/ ? Cheers, Thomas Hey Albert, (Actually I replied much earlier, but KMail refused to send the mail somehow) It's interesting that you can reproduce this on a Linux system, I never got that far. What version of kdelibs are you running? This particular issue should indeed be fixed in v4.12.2 and KF5. Actually it was only fixed for master (4.13) and KF5. I've now cherry-picked it to KDE/4.12 too since it fixes the crash i was getting. Thanks! Cheers, Albert Great, good to know! Cheers -- Kevin Funk
