Add the timestamp tick frequency within the timestamp table so
the cbmem utility doesn't try to figure it out on its own. Those
paths still exist for x86 systems which don't provide tsc_freq_mhz().
All other non-x86 systems use the monotonic timer which has a 1us
granularity or 1MHz.
One of the main reasons is that Linux is reporting
/sys/devices/system/cpu/cpu0/cpufreq/cpuinfo_max_freq as the true
turbo frequency on turbo enables machines. This change also fixes
the p-state values honored in cpufreq for turbo machines in that
turbo p-pstates were reported as 100MHz greater than nominal.
BUG=chrome-os-partner:44669
BRANCH=firmware-strago-7287.B
TEST=Built and booted on glados. Confirmed table frequency honored.
Change-Id: I763fe2d9a7b01d0ef5556e5abff36032062f5801
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: http://review.coreboot.org/11470
Tested-by: build bot (Jenkins)
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
Cherry-pick from chromium and adjusted for added boards
and changed directory layout for arch/arm.
Timestamp implementation for ARMv7
Abstract the use of rdtsc() and make the timestamps
uint64_t in the generic code.
The ARM implementation uses the monotonic timer.
Original-Signed-off-by: Stefan Reinauer <reinauer@google.com>
BRANCH=none
BUG=chrome-os-partner:18637
TEST=See cbmem print timestamps
Original-Change-Id: Id377ba570094c44e6895ae75f8d6578c8865ea62
Original-Reviewed-on: https://gerrit.chromium.org/gerrit/63793
(cherry-picked from commit cc1a75e059020a39146e25b9198b0d58aa03924c)
Change-Id: Ic51fb78ddd05ba81906d9c3b35043fa14fbbed75
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: http://review.coreboot.org/8020
Tested-by: build bot (Jenkins)
Reviewed-by: Marc Jones <marc.jones@se-eng.com>
The tsc header is using u32 w/o including the file
with defines it.
Change-Id: I9fcad882d25e93b4c0032b32abd2432b0169a068
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: http://review.coreboot.org/4843
Tested-by: build bot (Jenkins)
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
multiply_to_tsc was being copied everywhere, which is bad
practice. Put it in the tsc.h include file where it belongs.
Delete the copies of it.
Per secunet, no copyright notice is needed.
This might be a good time to get a copyright notice into tsc.h
anyway.
Change-Id: Ied0013ad4b1a9e5e2b330614bb867fd806f9a407
Signed-off-by: Ronald G. Minnich <rminnich@gmail.com>
Reviewed-on: http://review.coreboot.org/3242
Tested-by: build bot (Jenkins)
Reviewed-by: Nico Huber <nico.huber@secunet.com>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Some boards use the local apic for udelay(), but they also provide
their own implementation of udelay() for SMM. The reason for using
the local apic for udelay() in ramstage is to not have to pay the
penalty of calibrating the TSC frequency. Therefore provide a
TSC_CONSTANT_RATE option to indicate that TSC calibration is not
needed. Instead rely on the presence of a tsc_freq_mhz() function
provided by the cpu/board. Additionally, assume that if
TSC_CONSTANT_RATE is selected the udelay() function in SMM will
be the tsc.
Change-Id: I1629c2fbe3431772b4e80495160584fb6f599e9e
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: http://review.coreboot.org/3168
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
The CPU can arbitrarily reorder calls to rdtsc, significantly
reducing the precision of timing using the CPUs time stamp counter.
Unfortunately the method of synchronizing rdtsc is different
on AMD and Intel CPUs. There is a generic method, using the cpuid
instruction, but that uses up a lot of registers, and is very slow.
Hence, use the correct lfence/mfence instructions (for CPUs that
we know support it)
Change-Id: I17ecb48d283f38f23148c13159aceda704c64ea5
Signed-off-by: Stefan Reinauer <reinauer@google.com>
Reviewed-on: http://review.coreboot.org/1422
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Tested-by: build bot (Jenkins)
appropriate. Also, factor out post_code() for __PRE_RAM__ code and drop it from
some mainboards.
Signed-off-by: Stefan Reinauer <stepan@coresystems.de>
Acked-by: Stefan Reinauer <stepan@coresystems.de>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@5307 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
- drop include/part and move files to include/
- get rid lots of warnings
- make resource allocator happy with w83627thg
- trivial cbmem resume fix
- fix payload and log level settings in abuild
- fix kontron mptable for virtual wire mode
- drop some dead includes and dead code.
Signed-off-by: Stefan Reinauer <stepan@coresystems.de>
Acked-by: Ronald G. Minnich <rminnich@gmail.com>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@5136 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
__ROMCC__ now means "Don't use prototypes, since romcc doesn't support them."
__PRE_RAM__ means "Use simpler versions of functions, and no device tree."
There are probably some places where both are tested, but only one is needed.
Signed-off-by: Myles Watson <mylesgw@gmail.com>
Acked-by: Peter Stuge <peter@stuge.se>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@4921 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
TODO
- x86emu need (imo) some common header with prototypes at least
- clog2, ulzma, hardwaremain prototypes added by this patch probably should
be moved to some header too.
- in src/devices/device_util.c prototype is before function because seems,
it is used only within same file, if not it should be moved to debug
section of prototypes in include/device/device.h
Signed-off-by: Maciej Pijanka <maciej.pijanka@gmail.com>
Acked-by: Myles Watson <mylesgw@gmail.com>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@4871 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
