summaryrefslogtreecommitdiff
path: root/.clang-format
diff options
context:
space:
mode:
authorLinus Torvalds <torvalds@linux-foundation.org>2022-12-12 09:01:36 -0800
committerLinus Torvalds <torvalds@linux-foundation.org>2022-12-12 09:01:36 -0800
commit98d0052d0d9dcd5323833482712b5799ed0bbb0b (patch)
treee055e679d2aebf8b11d82b010775d96794295106 /.clang-format
parent73fa58dca80293320f5cfeb06f5b2daeb8d97bd5 (diff)
parent6b2b0d839acaa84f05a77184370f793752e786e9 (diff)
downloadlwn-98d0052d0d9dcd5323833482712b5799ed0bbb0b.tar.gz
lwn-98d0052d0d9dcd5323833482712b5799ed0bbb0b.zip
Merge tag 'printk-for-6.2' of git://git.kernel.org/pub/scm/linux/kernel/git/printk/linux
Pull printk updates from Petr Mladek: - Add NMI-safe SRCU reader API. It uses atomic_inc() instead of this_cpu_inc() on strong load-store architectures. - Introduce new console_list_lock to synchronize a manipulation of the list of registered consoles and their flags. This is a first step in removing the big-kernel-lock-like behavior of console_lock(). This semaphore still serializes console->write() calbacks against: - each other. It primary prevents potential races between early and proper console drivers using the same device. - suspend()/resume() callbacks and init() operations in some drivers. - various other operations in the tty/vt and framebufer susbsystems. It is likely that console_lock() serializes even operations that are not directly conflicting with the console->write() callbacks here. This is the most complicated big-kernel-lock aspect of the console_lock() that will be hard to untangle. - Introduce new console_srcu lock that is used to safely iterate and access the registered console drivers under SRCU read lock. This is a prerequisite for introducing atomic console drivers and console kthreads. It will reduce the complexity of serialization against normal consoles and console_lock(). Also it should remove the risk of deadlock during critical situations, like Oops or panic, when only atomic consoles are registered. - Check whether the console is registered instead of enabled on many locations. It was a historical leftover. - Cleanly force a preferred console in xenfb code instead of a dirty hack. - A lot of code and comment clean ups and improvements. * tag 'printk-for-6.2' of git://git.kernel.org/pub/scm/linux/kernel/git/printk/linux: (47 commits) printk: htmldocs: add missing description tty: serial: sh-sci: use setup() callback for early console printk: relieve console_lock of list synchronization duties tty: serial: kgdboc: use console_list_lock to trap exit tty: serial: kgdboc: synchronize tty_find_polling_driver() and register_console() tty: serial: kgdboc: use console_list_lock for list traversal tty: serial: kgdboc: use srcu console list iterator proc: consoles: use console_list_lock for list iteration tty: tty_io: use console_list_lock for list synchronization printk, xen: fbfront: create/use safe function for forcing preferred netconsole: avoid CON_ENABLED misuse to track registration usb: early: xhci-dbc: use console_is_registered() tty: serial: xilinx_uartps: use console_is_registered() tty: serial: samsung_tty: use console_is_registered() tty: serial: pic32_uart: use console_is_registered() tty: serial: earlycon: use console_is_registered() tty: hvc: use console_is_registered() efi: earlycon: use console_is_registered() tty: nfcon: use console_is_registered() serial_core: replace uart_console_enabled() with uart_console_registered() ...
Diffstat (limited to '.clang-format')
-rw-r--r--.clang-format1
1 files changed, 1 insertions, 0 deletions
diff --git a/.clang-format b/.clang-format
index 8d01225bfcb7..bcf60344f9f5 100644
--- a/.clang-format
+++ b/.clang-format
@@ -222,6 +222,7 @@ ForEachMacros:
- 'for_each_component_dais'
- 'for_each_component_dais_safe'
- 'for_each_console'
+ - 'for_each_console_srcu'
- 'for_each_cpu'
- 'for_each_cpu_and'
- 'for_each_cpu_not'