This allows the minimum time that default_delay() will choose to sleep
for instead of polling to be configured at build-time. The default
remains unchanged at 100 milliseconds for now.
The test's correctness has been checked by testing with minimum sleep
time left at its default and set to a non-default value smaller than 100
microseconds (both pass without sleeping, verified with strace) and with
the minimum sleep time set to 0 (causing the test to be skipped). The
configured value from the macro needs to be stored in a const to avoid
-Werror=type-limits errors when configured to be zero.
Change-Id: Ida96e0816ac914ed69d6fd82ad90ebe89cdef1cc
Signed-off-by: Peter Marheine <pmarheine@chromium.org>
Reviewed-on: https://review.coreboot.org/c/flashrom/+/81606
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Anastasia Klimchuk <aklm@chromium.org>
As proposed on the mailing list ("RFC: remove the calibrated delay
loop" [1]), this removes the calibrated delay loop and uses OS-based
timing functions for all delays because the calibrated delay loop can
delay for shorter times than intended.
When sleeping this now uses nanosleep() unconditionally, since usleep
was only used on DOS (where DJGPP lacks nanosleep). When busy-looping,
it uses clock_gettime() with CLOCK_MONOTONIC or CLOCK_REALTIME depending
on availability, and gettimeofday() otherwise.
The calibrated delay loop is retained for DOS only, because timer
resolution on DJGPP is only about 50 milliseconds. Since typical delays
in flashrom are around 10 microseconds, using OS timing there would
regress performance by around 500x. The old implementation is reused
with some branches removed based on the knowledge that timer resolution
will not be better than about 50 milliseconds.
Tested by reading and writing flash on several Intel and AMD systems:
* Lenovo P920 (Intel C620, read/verify only)
* "nissa" chromebook (Intel Alder Lake-N)
* "zork" chromebook (AMD Zen+)
[1]: https://mail.coreboot.org/hyperkitty/list/flashrom@flashrom.org/thread/HFH6UHPAKA4JDL4YKPSQPO72KXSSRGME/
Signed-off-by: Peter Marheine <pmarheine@chromium.org>
Change-Id: I7ac5450d194a475143698d65d64d8bcd2fd25e3f
Reviewed-on: https://review.coreboot.org/c/flashrom/+/81545
Reviewed-by: Anastasia Klimchuk <aklm@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Brian Norris <briannorris@chromium.org>
By default, we busy-loop (a.k.a., "delay") for most delay values, and
only allow sleeping for large delays. But busy-looping is expensive, as
it wastes CPU cycles.
In a simple program that runs a bunch of samples of [1] over 1000
samples, I find that for 0.1 s (100000 us):
64x2 AMD CPU (CONFIG_HZ=250 / CONFIG_NO_HZ_FULL=y):
min diff: 60 us
max diff: 831 us
mean diff: 135 us
4+2 Mediatek MT8183 CPU (CONFIG_HZ=1000 / CONFIG_NO_HZ_IDLE=y /
sysctl kernel.timer_highres=1):
min diff: 70 us
max diff: 1556 us
mean diff: 146 us
4+2 Mediatek MT8183 CPU (CONFIG_HZ=1000 / CONFIG_NO_HZ_IDLE=y /
sysctl kernel.timer_highres=0):
min diff: 94 us
max diff: 7222 us
mean diff: 1201 us
i.e., maximum 1.5% error, typically ~0.1% error with high resolution
timers. Max 7% error, typical 1% error with low resolution timers. The
error is always in the positive direction (i.e., sleep longer than the
requested delay, not shorter than the request).
This seems reasonable.
[1] Stripped / pseudocode:
clock_gettime(CLOCK_MONOTONIC, before);
nanosleep({ .tv_nsec = usecs * 1000 }, NULL);
clock_gettime(CLOCK_MONOTONIC, after);
diff = abs((after - before) / 1000 - usecs));
Change-Id: Ifd4821c66c5564f7c975c08769a6742f645e9be0
Signed-off-by: Brian Norris <briannorris@chromium.org>
Reviewed-on: https://review.coreboot.org/c/flashrom/+/80808
Reviewed-by: Anastasia Klimchuk <aklm@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
The non-custom driver programmer delay implementation
'internal_delay()' is unrelated specifically to the
'internal' programmer. The delay implementation is
simply a platform-agnostic host delay implementation.
Therefore, rename to simply default_delay().
Change-Id: I5e04adf16812ceb1480992c92bca25ed80f8897a
Signed-off-by: Edward O'Callaghan <quasisec@google.com>
Reviewed-on: https://review.coreboot.org/c/flashrom/+/68855
Reviewed-by: Alexander Goncharov <chat@joursoir.net>
Reviewed-by: Felix Singer <felixsinger@posteo.net>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
These files all contain functions whose prototypes are in header files,
so make sure those header files are included.
Change-Id: I0189a1550bf90d4a0b87dcef9f8a8449590cc9d7
Signed-off-by: Jacob Garber <jgarber1@ualberta.ca>
Reviewed-on: https://review.coreboot.org/c/flashrom/+/33668
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Nico Huber <nico.h@gmx.de>
Instead of calibrating our busy loop against a coarse clock, check if
a precise clock is available and loop against that. The former is unre-
liable by definition on any modern system that may dynamically reclock
the processor.
v2: Apparently _POSIX_MONOTONIC_CLOCK being defined only means that
the library knows about CLOCK_MONOTONIC. So check for its support
at runtime and fall back to CLOCK_REALTIME if it's missing.
TEST=Manually added a 10s loop and compared to real time. Run
on Linux RPi3, Linux x86 and my original use case Linux in
VirtualBox (Linux host).
Change-Id: I85ad359823875237ada9cd027af3017d62e9a235
Signed-off-by: Nico Huber <nico.h@gmx.de>
Reviewed-on: https://review.coreboot.org/19391
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: David Hendricks <david.hendricks@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Usleep() has been obsolete for quite a while.
The only target that uses it without alternative is DOS.
Corresponding to flashrom svn r1899.
Signed-off-by: Stefan Tauner <stefan.tauner@alumni.tuwien.ac.at>
Acked-by: Stefan Tauner <stefan.tauner@alumni.tuwien.ac.at>
We do CPU architecture checks once for the makefile in arch.h and
once for HW access abstraction in hwaccess.c. This patch unifies
related files so that they can share the checks to improve
maintainability and reduce the chance of inconsistencies.
Furthermore, it refines some of the definitions, which
- adds "support" for AARCH64 and PPC64,
- adds big-endian handling on arm as well as LE handling on PPC64,
- fixes compilation of internal.c on AARCH64 and PPC64.
Additionally, this patch continues to unify all OS checks in
flashrom by adding a new helper macro IS_WINDOWS.
The old header file for architecture checking is renamed to platform.h
to reflect its broader scope and all new macros are add in there.
Corresponding to flashrom svn r1864.
Signed-off-by: Stefan Tauner <stefan.tauner@alumni.tuwien.ac.at>
Acked-by: Stefan Tauner <stefan.tauner@alumni.tuwien.ac.at>
There is no reason for negative delays in our use cases:
- We don't need it (to work around any quirks).
- sleep() (POSIX) uses an unsigned argument.
- usleep() (POSIX) uses an unsigned argument.
- Sleep() (Windows) uses an unsigned argument.
Change all callees as well (without any complications).
Corresponding to flashrom svn r1782.
Signed-off-by: Stefan Tauner <stefan.tauner@alumni.tuwien.ac.at>
Acked-by: Stefan Tauner <stefan.tauner@alumni.tuwien.ac.at>
Handle long sleeps on non-Windows correctly.
Corresponding to flashrom svn r1667.
Signed-off-by: Maksim Kuleshov <mmcx@mail.ru>
Signed-off-by: Carl-Daniel Hailfinger <c-d.hailfinger.devel.2006@gmx.net>
Acked-by: Stefan Tauner <stefan.tauner@alumni.tuwien.ac.at>
This patch just fixes a limited number of bits not conforming to c99 by using
- __asm__ instead of just asm
- {0} instead of {} for struct initialization
- h_addr_list[0] instead of h_addr to access the host address in
struct hostent
- #include <strings.h> where needed (for ffs and strcasecmp)
Based on a previous patch by Carl-Daniel.
Corresponding to flashrom svn r1585.
Signed-off-by: Carl-Daniel Hailfinger <c-d.hailfinger.devel.2006@gmx.net>
Signed-off-by: Stefan Tauner <stefan.tauner@alumni.tuwien.ac.at>
Acked-by: Carl-Daniel Hailfinger <c-d.hailfinger.devel.2006@gmx.net>
- Fix various minor compile issues (eg. include necessary standard headers)
- Fix compilation of libpayload code paths
- Provide libpayload support in Makefile
- Add make target "libflashrom.a" which links non-CLI code to static
library
Corresponding to flashrom svn r1280.
Signed-off-by: Patrick Georgi <patrick.georgi@secunet.com>
Tested-with-DOS-crosscompiler-by: Idwer Vollering <vidwer@gmail.com>
Acked-by: Carl-Daniel Hailfinger <c-d.hailfinger.devel.2006@gmx.net>
This patch fixes the "using plain integer as NULL pointer" warnings
generated by running sparse on the flashrom source.
Corresponding to flashrom svn r1255.
Signed-off-by: Peter Huewe <peterhuewe@gmx.de>
Acked-by: Mathias Krause <mathias.krause@secunet.com>
Acked-by: Stefan Reinauer <stepan@coreboot.org>
This doesn't include changes to the frontend which must be
done separately, so this won't work out of the box.
This code was tested on hardware.
Corresponding to flashrom svn r1184.
Signed-off-by: Patrick Georgi <patrick.georgi@coresystems.de>
Acked-by: Carl-Daniel Hailfinger <c-d.hailfinger.devel.2006@gmx.net>
Constify variables where possible.
Initialize programmer-related variables explicitly in programmer_init to
allow running programmer_init from a clean state after
programmer_shutdown.
Prohibit registering programmer shutdown functions before init or after
shutdown.
Kill some dead code.
Rename global variables with namespace-polluting names.
Use a previously unused locking helper function in sst49lfxxxc.c.
This is needed for libflashrom.
Effects on the binary size of flashrom are minimal (300 bytes
shrinkage), but the data section shrinks by 4384 bytes, and that's a
good thing if flashrom is operating in constrained envionments.
Corresponding to flashrom svn r1068.
Signed-off-by: Carl-Daniel Hailfinger <c-d.hailfinger.devel.2006@gmx.net>
Acked-by: Michael Karcher <flashrom@mkarcher.dialup.fu-berlin.de>
Since we don't support any other compilers right now, enable that
extension unconditionally.
Fix the bugs found by format string checking.
Corresponding to flashrom svn r1032.
Signed-off-by: Carl-Daniel Hailfinger <c-d.hailfinger.devel.2006@gmx.net>
Acked-by: Sean Nelson <audiohacked@gmail.com>
Do not trust the OS at all and measure timer precision before
calibrating the delay loop and use that measurement to get reasonable
precision for our own delay code.
Print a measurement for a delay of 4x the OS timer resolution.
Be precise about how bad the deviation was if we had to recalculate.
Tested on Windows XP, 32 bit, built using MinGW by Uwe. Tested on
FreeDOS v1.0 Final, 32bit, built using DJGPP 4.3.2 by Idwer.
Corresponding to flashrom svn r1028.
Signed-off-by: Carl-Daniel Hailfinger <c-d.hailfinger.devel.2006@gmx.net>
Acked-by: Uwe Hermann <uwe@hermann-uwe.de>
Acked-by: Idwer Vollering <vidwer+lists.flashrom@gmail.com>
Unistd.h was only used to get a definition of NULL in all files. Add our
own NULL #define and remove unistd.h from flash.h
stdio.h has no place in flash.h, it should be included only in files
which really need it.
Add #include statements in individual .c files where needed.
Replace a few printf with msg_* to eliminate the need for stdio.h.
Corresponding to flashrom svn r1021.
Signed-off-by: Carl-Daniel Hailfinger <c-d.hailfinger.devel.2006@gmx.net>
Acked-by: Uwe Hermann <uwe@hermann-uwe.de>
Fix an unescaped % in a format specifier.
Corresponding to flashrom svn r992.
Signed-off-by: Urja Rannikko <urjaman@gmail.com>
Acked-by: Carl-Daniel Hailfinger <c-d.hailfinger.devel.2006@gmx.net>
The current delay loop calculation is still from revision 1 of flashrom,
and since then it had a logic bug which caused all delays to be twice as
long as intended.
Fix the delay duration.
Protect against delay loop overflows.
Detect a non-working delay loop.
Change the delay loop itself to ensure clever compiler optimizers won't
eliminate it (as happens with clang/llvm in the current code). Some
people suggested machine-specific asm, but the empty asm statement with
the loop counter as register/memory input has the benefit of being
perfectly cross-platform and working in gcc and clang.
If time goes backwards (catastrophical NTP time difference, manual
time change), timing measurements were shot because the new-old time
subtraction yielded negative numbers which weren't handled correctly
because the variable is unsigned. Work around that issue (a fix is
mathematically impossible).
If time goes forward too fast, pick the biggest possible timing
measurement with a guaranteed overflow avoidance for all timing
calculations.
Check four times if the calculated timing is at most 10% too fast. This
addresses OS scheduler interactions, e.g. being scheduled out during
measurement which inflates measurements.
If the timing looks like garbage, recalculate the timer values up to
four times before giving up.
Avoid division by zero in rare cases where timing measurements for a 250
ms delay returned 0 us elapsed.
Corresponding to flashrom svn r990.
Signed-off-by: Carl-Daniel Hailfinger <c-d.hailfinger.devel.2006@gmx.net>
Acked-by: Maciej Pijanka <maciej.pijanka@gmail.com>
Clean up code duplication and measure timing of 10/100/1000/10000 us
delays.
Corresponding to flashrom svn r986.
Signed-off-by: Carl-Daniel Hailfinger <c-d.hailfinger.devel.2006@gmx.net>
Acked-by: Maciej Pijanka <maciej.pijanka@gmail.com>
Move that function to udelay.c and compile internal.c only if really
needed. physmap.c is only needed if the programmer is internal or a PCI
card. Make its compilation conditional.
Corresponding to flashrom svn r822.
Signed-off-by: Carl-Daniel Hailfinger <c-d.hailfinger.devel.2006@gmx.net>
Acked-by: Idwer Vollering <vidwer@gmail.com>
Add external programmer delay functions so external programmers can
handle the delay on their own if needed.
Corresponding to flashrom svn r578.
Signed-off-by: Carl-Daniel Hailfinger <c-d.hailfinger.devel.2006@gmx.net>
Acked-by: Uwe Hermann <uwe@hermann-uwe.de>
Acked-by: Urja Rannikko <urjaman@gmail.com>
Build-tested on 32bit x86.
Corresponding to flashrom svn r521.
Signed-off-by: Uwe Hermann <uwe@hermann-uwe.de>
Acked-by: Uwe Hermann <uwe@hermann-uwe.de>
Corresponding to flashrom svn r429 and coreboot v2 svn r4117.
Signed-off-by: Uwe Hermann <uwe@hermann-uwe.de>
Acked-by: Uwe Hermann <uwe@hermann-uwe.de>
Corresponding to flashrom svn r326 and coreboot v2 svn r3669.
Signed-off-by: Uwe Hermann <uwe@hermann-uwe.de>
Acked-by: Uwe Hermann <uwe@hermann-uwe.de>
Corresponding to flashrom svn r151 and coreboot v2 svn r2873.
Signed-off-by: Uwe Hermann <uwe@hermann-uwe.de>
Acked-by: Stefan Reinauer <stepan@coresystems.de>
Corresponding to flashrom svn r143 and coreboot v2 svn r2847.
Signed-off-by: Uwe Hermann <uwe@hermann-uwe.de>
Acked-by: Uwe Hermann <uwe@hermann-uwe.de>
Corresponding to flashrom svn r142 and coreboot v2 svn r2846.
Signed-off-by: Uwe Hermann <uwe@hermann-uwe.de>
Acked-by: Uwe Hermann <uwe@hermann-uwe.de>
Corresponding to flashrom svn r136 and coreboot v2 svn r2768.
Signed-off-by: Uwe Hermann <uwe@hermann-uwe.de>
Acked-by: Uwe Hermann <uwe@hermann-uwe.de>
I'm self-ack'ing this, as the origin of the code in udelay.c (and thus
the license and copyright owner) is pretty clear.
The code which is now in udelay.c was split out from flash_rom.c in r1428,
and flash_rom.c, in turn, has been around since the beginning and had a
'Copyright 2000 Silicon Integrated System Corporation' line as well as the
usual GPLv2-or-later license header.
Corresponding to flashrom svn r135 and coreboot v2 svn r2767.
Signed-off-by: Uwe Hermann <uwe@hermann-uwe.de>
Acked-by: Uwe Hermann <uwe@hermann-uwe.de>
Corresponding to flashrom svn r128 and coreboot v2 svn r2746.
Signed-off-by: Uwe Hermann <uwe@hermann-uwe.de>
Acked-by: Stefan Reinauer <stepan@coresystems.de>
* Only open /dev/mem once and do it early.
* Drop extern for function prototypes.
* Minimize ts5300 impact in probe_flash()
This cleanup will making ICH7 SPI support quite some easier.
Corresponding to flashrom svn r100 and coreboot v2 svn r2585.
Signed-off-by: Stefan Reinauer <stepan@coresystems.de>
Acked-by: Stefan Reinauer <stepan@coresystems.de>
Corresponding to flashrom svn r97 and coreboot v2 svn r2577.
Signed-off-by: Uwe Hermann <uwe@hermann-uwe.de>
Acked-by: Uwe Hermann <uwe@hermann-uwe.de>
