#
95ee2897 |
|
16-Aug-2023 |
Warner Losh <imp@FreeBSD.org> |
sys: Remove $FreeBSD$: two-line .h pattern Remove /^\s*\*\n \*\s+\$FreeBSD\$$\n/
|
#
1e4b7ab1 |
|
04-Jun-2023 |
Colin Percival <cperciva@FreeBSD.org> |
tslog.h: Wrap in #ifdef _KERNEL This is not intended to be used from outside of the kernel; in particular, the boot loader has its own version of tslog.
|
#
46dd801a |
|
16-Oct-2021 |
Colin Percival <cperciva@FreeBSD.org> |
Add userland boot profiling to TSLOG On kernels compiled with 'options TSLOG', record for each process ID: * The timestamp of the fork() which creates it and the parent process ID, * The first path passed to execve(), if any, * The first path resolved by namei, if any, and * The timestamp of the exit() which terminates the process. Expose this information via a new sysctl, debug.tslog_user. On kernels lacking 'options TSLOG' (the default), no information is recorded and the sysctl does not exist. Note that recording namei is needed in order to obtain the names of rc.d scripts being launched, as the rc system sources them in a subshell rather than execing the scripts. With this commit it is now possible to generate flamecharts of the entire boot process from the start of the loader to the end of /etc/rc. The code needed to perform this processing is currently found in github: https://github.com/cperciva/freebsd-boot-profiling Reviewed by: mhorne Sponsored by: https://www.patreon.com/cperciva Differential Revision: https://reviews.freebsd.org/D32493
|
#
0337438d |
|
31-Dec-2017 |
Colin Percival <cperciva@FreeBSD.org> |
Wrap includes in sys/tslog.h with #ifdef TSLOG. This is necessary because some non-kernel code #defines _KERNEL and then includes kernel headers; as a result, it was getting conflicting versions of curthread and curproc. Non-kernel code should probably refrain from defining _KERNEL, but for now hiding these indirect inclusions fixes the build. Reported by: Michael Butler, Herbert J. Skuhra
|
#
e31e7199 |
|
31-Dec-2017 |
Colin Percival <cperciva@FreeBSD.org> |
Code for recording timestamps of events, especially function entries/exits. This is a very primitive system, intended for use in measuring performance during the early system boot, before more sophisticated tools like DTrace or infrastructure like kernel memory allocation and mutexes are available. Because this code records pointers to strings rather than copying strings (in order to keep the memory usage more manageable), if a kernel module is unloaded after logging an event, Bad Things can happen. Users are advised to not do that. Since cycle counts from the early kernel boot are used as an initial entropy source, publishing this information to userland could result in inadequate entropy being kept private to the kernel RNG. Users are advised to not enable this on systems with untrusted users. Discussed on: freebsd-current
|