Skip to content

K
Kernel

openeuler / Kernel
11 个月 前同步成功

通知 5
Star 0
Fork 0
K
Kernel

体验新版 GitCode，发现更多精彩内容 >>

切换分支/标签
  1. 23 4月, 2008 1 次提交
  3. 22 4月, 2008 1 次提交
  5. 20 4月, 2008 1 次提交
  7. 18 4月, 2008 2 次提交
    • J
      kgdb: allow static kgdbts boot configuration · 974460c5
      由 Jason Wessel 提交于 
      This patch adds in the ability to compile the kgdb internal test
string into the kernel so as to run the tests at boot without changing
the kernel boot arguments.  This patch also changes all the error
paths to invoke WARN_ON(1) which will emit the line number of the file
and dump the kernel stack when an error occurs.

You can disable the tests in a kernel that is built this way
using "kgdbts="
Signed-off-by: NJason Wessel <jason.wessel@windriver.com>
Signed-off-by: NIngo Molnar <mingo@elte.hu>
      974460c5
    • J
      kgdb: add kgdb internal test suite · e8d31c20
      由 Jason Wessel 提交于 
      This patch adds regression tests for testing the kgdb core and arch
specific implementation.

The kgdb test suite is designed to be built into the kernel and not as
a module because it uses a number of low level kernel and kgdb
primitives which should not be exported externally.

The kgdb test suite is designed as a KGDB I/O module which
simulates the communications that a debugger would have with kgdb.
The tests are broken up in to a line by line and referenced here as
a "get" which is kgdb requesting input and "put" which is kgdb
sending a response.

The kgdb suite can be invoked from the kernel command line
arguments system or executed dynamically at run time.  The test
suite uses the variable "kgdbts" to obtain the information about
which tests to run and to configure the verbosity level.  The
following are the various characters you can use with the kgdbts=
line:

When using the "kgdbts=" you only choose one of the following core
test types:
A = Run all the core tests silently
V1 = Run all the core tests with minimal output
V2 = Run all the core tests in debug mode

You can also specify optional tests:
N## = Go to sleep with interrupts of for ## seconds
      to test the HW NMI watchdog
F## = Break at do_fork for ## iterations
S## = Break at sys_open for ## iterations

NOTE: that the do_fork and sys_open tests are mutually exclusive.

To invoke the kgdb test suite from boot you use a kernel start
argument as follows:
	kgdbts=V1 kgdbwait
Or if you wanted to perform the NMI test for 6 seconds and do_fork
test for 100 forks, you could use:
	kgdbts=V1N6F100 kgdbwait

The test suite can also be invoked at run time with:
echo kgdbts=V1N6F100 > /sys/module/kgdbts/parameters/kgdbts
Or as another example:
echo kgdbts=V2 > /sys/module/kgdbts/parameters/kgdbts

When developing a new kgdb arch specific implementation or
using these tests for the purpose of regression testing,
several invocations are required.

1) Boot with the test suite enabled by using the kernel arguments
      "kgdbts=V1F100 kgdbwait"
   ## If kgdb arch specific implementation has NMI use
      "kgdbts=V1N6F100

2) After the system boot run the basic test.
echo kgdbts=V1 > /sys/module/kgdbts/parameters/kgdbts

3) Run the concurrency tests.  It is best to use n+1
   while loops where n is the number of cpus you have
   in your system.  The example below uses only two
   loops.

## This tests break points on sys_open
while [ 1 ] ; do find / > /dev/null 2>&1 ; done &
while [ 1 ] ; do find / > /dev/null 2>&1 ; done &
echo kgdbts=V1S10000 > /sys/module/kgdbts/parameters/kgdbts
fg # and hit control-c
fg # and hit control-c
## This tests break points on do_fork
while [ 1 ] ; do date > /dev/null ; done &
while [ 1 ] ; do date > /dev/null ; done &
echo kgdbts=V1F1000 > /sys/module/kgdbts/parameters/kgdbts
fg # and hit control-c
Signed-off-by: NJason Wessel <jason.wessel@windriver.com>
Signed-off-by: NIngo Molnar <mingo@elte.hu>
      e8d31c20
  9. 16 4月, 2008 1 次提交
  11. 31 3月, 2008 1 次提交
  13. 13 3月, 2008 3 次提交
  15. 12 3月, 2008 2 次提交
  17. 11 3月, 2008 1 次提交
  19. 05 3月, 2008 1 次提交
  21. 04 3月, 2008 2 次提交
    • D
      atmel_tc clocksource/clockevent code · 4d243f92
      由 David Brownell 提交于 
      Clocksource and clockevent device based on the Atmel TC blocks.

The clockevent device handles both periodic and oneshot modes, so this
enables NO_HZ and high res timers on some platforms that previously
couldn't use those mechanisms.

This works on both AVR32 and AT91 chips, given relevant patches for
tclib support (always) and clockevents (or else this will only look
like a higher precision clocksource).  It's an updated and modularized
version of an AT91-only patch that has circulated for some time now.

Changes relative to the original patch:
  * Update to use new tclib API
  * Replace open-coded do-while loop using goto with a real do-while loop
  * Minor irq handler optimization: Load register base address from
    dev_id instead of a global variable.
  * Aggressively turn off clocks when the clockevent isn't being used
  * Include the clockevent code on AT91RM9200 as well. The rating is
    lower than the System Timer, so the clock will usually stay off.
  * Don't assume that the number of clocks is always equal to the
    number of irqs.
Signed-off-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: NHaavard Skinnemoen <hskinnemoen@atmel.com>
      4d243f92
    • D
      atmel_tc library · 2a341f5c
      由 David Brownell 提交于 
      Create <linux/atmel_tc.h> based on <asm-arm/arch-at91/at91-tc.h> and the
at91sam9263 and at32ap7000 datasheets.  Most AT91 and AT32 SOCs have one
or two of these TC blocks, which include three 16-bit timers that can be
interconnected in various ways.

These TC blocks can be used for external interfacing (such as PWM and
measurement), or used as somewhat quirky sixteen-bit timers.

Changes relative to the original version:
  * Drop unneeded inclusion of <linux/mutex.h>
  * Support an arbitrary number of TC blocks
  * Return a struct with information about a TC block from
    atmel_tc_alloc() instead of using a combination of return values
    and "out" parameters.
  * ioremap() the I/O registers on allocation
  * Look up clocks and irqs for all channels
  * Add "name" parameter to atmel_tc_alloc() and use this when
    requesting the iomem resource.
  * Check if the platform provided the necessary resources at probe()
    time instead of when the TCB is allocated.
Signed-off-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: NHaavard Skinnemoen <hskinnemoen@atmel.com>
      2a341f5c
  23. 21 2月, 2008 2 次提交
    • A
      sony-laptop.c: fix off-by-one · d399d130
      由 Adrian Bunk 提交于 
      This patch fixes an off-by-one spotted by the Coverity checker.
Signed-off-by: NAdrian Bunk <bunk@kernel.org>
Acked-by: NMattia Dongili <malattia@linux.it>
Signed-off-by: NLen Brown <len.brown@intel.com>
      d399d130
    • S
      acer-wmi: fix section mismatch warnings · 7560e385
      由 Sam Ravnborg 提交于 
      Fix following warnings:
WARNING: vmlinux.o(.text+0x672615): Section mismatch in reference from the function acer_platform_remove() to the function .exit.text:acer_backlight_exit()
WARNING: vmlinux.o(.devinit.text+0x1e859): Section mismatch in reference from the function acer_platform_probe() to the function .init.text:acer_led_init()
WARNING: vmlinux.o(.devinit.text+0x1e878): Section mismatch in reference from the function acer_platform_probe() to the function .init.text:acer_backlight_init()

Remove __exit annotation from acer_backlight_exit(). We cannot reference
a __exit annotated function from non __exit functions.

acer_led_init() and acer_backlight_init() where both annotated __init but
used from a __devinit function. This would result in an oops should
gcc drop their inlining and the module are hot plugged.

Fix by annotating acer_led_init() and acer_backlight_init() __devinit.
Signed-off-by: NSam Ravnborg <sam@ravnborg.org>
Cc: Carlos Corbacho <carlos@strangeworlds.co.uk>
Signed-off-by: NLen Brown <len.brown@intel.com>
      7560e385
  25. 16 2月, 2008 9 次提交
  27. 14 2月, 2008 1 次提交
  29. 10 2月, 2008 1 次提交
  31. 09 2月, 2008 6 次提交
  33. 08 2月, 2008 3 次提交
  35. 07 2月, 2008 2 次提交