Patrick J Volkerding 9d31c53bf5 Sat Oct 23 18:57:30 UTC 2021 ... a/aaa_terminfo-6.3-x86_64-1.txz: Upgraded. a/glibc-zoneinfo-2021e-noarch-1.txz: Upgraded. ap/itstool-2.0.7-x86_64-2.txz: Rebuilt. Rebuilt with PYTHON=/usr/bin/python3. Thanks to USUARIONUEVO. ap/mpg123-1.29.2-x86_64-1.txz: Upgraded. d/meson-0.59.3-x86_64-1.txz: Upgraded. d/parallel-20211022-noarch-1.txz: Upgraded. d/python-pip-21.3.1-x86_64-1.txz: Upgraded. d/python-setuptools-58.3.0-x86_64-1.txz: Upgraded. l/exiv2-0.27.5-x86_64-1.txz: Upgraded. l/ncurses-6.3-x86_64-1.txz: Upgraded. n/php-7.4.25-x86_64-1.txz: Upgraded. This update fixes bugs and a security issue: FPM: PHP-FPM oob R/W in root process leading to privilege escalation. For more information, see: https://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2021-21703 (* Security fix *) xap/mozilla-thunderbird-91.2.1-x86_64-1.txz: Upgraded. This is a bugfix release. For more information, see: https://www.mozilla.org/en-US/thunderbird/91.2.1/releasenotes/ testing/packages/linux-5.14.x/kernel-generic-5.14.14-x86_64-2.txz: Rebuilt. testing/packages/linux-5.14.x/kernel-headers-5.14.14-x86-2.txz: Rebuilt. testing/packages/linux-5.14.x/kernel-huge-5.14.14-x86_64-2.txz: Rebuilt. testing/packages/linux-5.14.x/kernel-modules-5.14.14-x86_64-2.txz: Rebuilt. testing/packages/linux-5.14.x/kernel-source-5.14.14-noarch-2.txz: Rebuilt. These kernels enable CONFIG_PREEMPT=y and CONFIG_PREEMPT_DYNAMIC=y allowing the kernel preemption model to be specified on the kernel command line with one of these options: preempt=none, preempt=voluntary, and preempt=full. Since there is no .config option to set a default, and the default in the kernel sources is "full" (which is probably not a good default), the kernel-source.SlackBuild has been modified to add support for an environment variable CONFIG_PREEMPT_DEFAULT_MODE which can be set to none, voluntary, or full to set the default kernel preemption model when a command line option is not provided. These kernels have been built with a preemption model of "none" (presumably the safest choice which will behave like the kernels we have shipped before.) The runtime overhead on 64-bit should be negligible. On 32-bit we lack support for HAVE_STATIC_CALL_INLINE, so spinlocks and mutexes will have to be approached through a trampoline, adding a very small amount of overhead. I feel this is probably worth it in order to have the option to run a kernel with voluntary or full preemption, especially for gaming or desktop purposes. The reduction in input lag with these modes is actually quite noticable. To check the current preemption model, you may use debugfs: mount -t debugfs none /sys/kernel/debug cat /sys/kernel/debug/sched/preempt (none) voluntary full You may change to a different preemption model on the fly once debugfs is mounted: echo voluntary > /sys/kernel/debug/sched/preempt cat /sys/kernel/debug/sched/preempt none (voluntary) full Thanks to Daedra. -DRM_I810 n -INLINE_READ_UNLOCK y -INLINE_READ_UNLOCK_IRQ y -INLINE_SPIN_UNLOCK_IRQ y -INLINE_WRITE_UNLOCK y -INLINE_WRITE_UNLOCK_IRQ y PREEMPT n -> y PREEMPT_VOLUNTARY y -> n +CEC_GPIO n +DEBUG_PREEMPT y +PREEMPTION y +PREEMPT_COUNT y +PREEMPT_DYNAMIC y +PREEMPT_RCU y +PREEMPT_TRACER n +RCU_BOOST n +TASKS_RCU y +UNINLINE_SPIN_UNLOCK y		2021-10-24 08:59:56 +02:00
..
itstool.SlackBuild	Sat Oct 23 18:57:30 UTC 2021	2021-10-24 08:59:56 +02:00
slack-desc	Mon May 28 19:12:29 UTC 2018	2018-05-31 23:39:35 +02:00