We have converted all LPC-only configs to HOSTCMD_LPC so the remaining
CONFIG_LPC defines represent the common case.
BRANCH=none
BUG=chromium:818804
TEST=Full stack builds and works on yorp (espi) and grunt (lpc)
Change-Id: Iba9a48f2cab12fadd0d9ab8eab0d5d5476eab238
Signed-off-by: Jett Rink <jettrink@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/1067503
Reviewed-by: Randall Spangler <rspangler@chromium.org>
Break the ec chip code up with the more granular
CONFIG_HOSTCMD_(X86|LPC|ESPI) options.
BRANCH=none
BUG=chromium:818804
TEST=Full stack builds and works on yorp (espi) and grunt (lpc)
Change-Id: Ie272787b2425175fe36b06fcdeeee90ec5ccbe95
Signed-off-by: Jett Rink <jettrink@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/1067502
Reviewed-by: Randall Spangler <rspangler@chromium.org>
gcc 8.1 complains about duplicate const, and while some of these really
are duplicate, others look like they were supposed to tighten the API
contract so that variables are "const pointer to const data", but didn't
have that effect.
BUG=b:65441143
BRANCH=none
TEST=building Chrome EC as part of upstream coreboot's build with a
gcc 8.1 compiler now works (better. there are other issues left)
Change-Id: I6016c5f282516471746f08d5714ea07ebdd10331
Signed-off-by: Patrick Georgi <pgeorgi@google.com>
Reviewed-on: https://chromium-review.googlesource.com/1039812
Commit-Ready: Patrick Georgi <pgeorgi@chromium.org>
Tested-by: Patrick Georgi <pgeorgi@chromium.org>
Reviewed-by: Stefan Reinauer <reinauer@google.com>
Reviewed-by: Vincent Palatin <vpalatin@chromium.org>
They're only used within the same file and should always be inlined.
It also helps gcc 8.1's lto linking which seems to not inline it (since
inline is just a hint) but then drops the function (presumably because
it's small, marked inline, and comes with no prototype).
BUG=b:65441143
BRANCH=none
TEST=builds with gcc 8.1
Change-Id: I881a5b9f13192dd11748d8a3060788f95a84dec0
Signed-off-by: Patrick Georgi <pgeorgi@google.com>
Reviewed-on: https://chromium-review.googlesource.com/1061075
Commit-Ready: Patrick Georgi <pgeorgi@chromium.org>
Tested-by: Patrick Georgi <pgeorgi@chromium.org>
Reviewed-by: Stefan Reinauer <reinauer@google.com>
Sometimes common code needs __hw_clock_source_read, add it.
The implementation is similar of what common/timer.c does to create a ts
for get_time(), but in reverse.
TEST=Unit tests pass again for the next CL
BRANCH=master
BUG=None
Change-Id: I10564abedabe88e4789723bc97bac170ae020c69
Signed-off-by: Alexandru M Stan <amstan@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/1055191
Reviewed-by: Vincent Palatin <vpalatin@chromium.org>
We add a configuration option to set the minimum shared memory
size (CONFIG_SHAREDMEM_MINIMUM_SIZE), so that the link will fail
if there is not enough IRAM left.
Also, we add 2 macros around shared_mem_acquire, that check, at
build time, that the shared memory size is sufficient for the
allocation:
- SHARED_MEM_ACQUIRE_CHECK should be used instead of
shared_mem_acquire, when size is known in advance.
- SHARED_MEM_CHECK_SIZE should be used when only a maximum size
is known.
This does not account for "jump tags" that boards often add on
jump from RO to RW. Luckily, RW usually does not do verification,
and does not need as much shared memory.
BRANCH=none
BUG=chromium:739771
TEST=make buildall -j, no error
Change-Id: Ic4c72938affe65fe8f8bc17ee5111c1798fc536f
Signed-off-by: Nicolas Boichat <drinkcat@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/1002713
Reviewed-by: Gwendal Grignou <gwendal@chromium.org>
Reviewed-by: Randall Spangler <rspangler@chromium.org>
Reviewed-by: Vincent Palatin <vpalatin@chromium.org>
We got a symptom that keyboard didn't work without connecting
servo board after this change (CL:897315) was merged.
This is because our uart RX will receive a data (0) with framing
error if RX level is low and trigger a re-scheduling request in
uart ISR (HOOKS task will be wake at later and then start the
task scheduling).
And that will cause we don't return to main() function to finish
all operations of initialization after uart_init() is called.
I think we will get the same symptom if GPIO/peripheral interrupts
are enabled and wake some task at initialization.
This change makes sure we will start the task scheduling if
task_start() is called.
BUG=none
BRANCH=none
TEST=With this change, keyboard works after EC reboot without
servo board connected.
Change-Id: I0bda84b1cb56ced6aad2a38b0786d1b336e77211
Signed-off-by: Dino Li <Dino.Li@ite.com.tw>
Reviewed-on: https://chromium-review.googlesource.com/956794
Reviewed-by: Vincent Palatin <vpalatin@chromium.org>
Remove the former special case for USB RAM
Add additional RAM regions for STM32H7.
For USB RAM, add an explicit alignment directive to ensure we always meet
the 8-byte boundary hardware constraint for the BTABLE.
This was already true because we put the .usb_ram.btable section first.
I keep this property by alpha-sorting the sections but makes it more
explicit by adding a 2-digit numeric prefix: e.g. 00_firstsection,
99_lastsection.
Signed-off-by: Vincent Palatin <vpalatin@chromium.org>
BRANCH=none
BUG=b:67081508
TEST=on ZerbleBarn, along with the following CLs, run the firmware with
large arrays in special AHB memory regions.
TEST=build all targets with and without the patch and verify that all
smap files are identical.
Change-Id: I9ee7f519a13cb14ba9997220f22180028f9c0175
Reviewed-on: https://chromium-review.googlesource.com/946369
Commit-Ready: Vincent Palatin <vpalatin@chromium.org>
Tested-by: Nicolas Boichat <drinkcat@chromium.org>
Reviewed-by: Nicolas Boichat <drinkcat@chromium.org>
When a chip has special/non-contiguous SRAM physical memory region,
rather than extending the generic linker file ad nauseam, define a
mechanism to declare a chip specific list of those regions.
To do so, a chip must declare the CONFIG_CHIP_MEMORY_REGIONS
configuration and have a memory_regions.inc with the list of regions.
The special-purpose preprocessed chip/<chip_name>/memory_regions.inc
file has one region declaration per line using the following macro:
REGION(name, attributes, start_address, size)
Each region will get a proper MEMORY entry and a section in the linker
file.
the __SECTION(region_name) helper is provided as a convenience to
declare variable in a specific region.
Note: those 'special' regions are NOT cleared at startup contrary to
.bss.
Signed-off-by: Vincent Palatin <vpalatin@chromium.org>
BRANCH=none
BUG=b:67081508
TEST=on ZerbleBarn, along with the following CLs, run the firmware with
large arrays in special AHB memory regions.
Change-Id: I3f156ef6e5feb4a6a0b2ae2468bae8a20483f17c
Reviewed-on: https://chromium-review.googlesource.com/946368
Commit-Ready: Vincent Palatin <vpalatin@chromium.org>
Tested-by: Vincent Palatin <vpalatin@chromium.org>
Reviewed-by: Randall Spangler <rspangler@chromium.org>
Cr50 lacks native instructions for 64-bit integers and an ABI
function can be used by the compiler to take the place of the
needed instructions. This CL adds support for a right bitwise
shift of 64-bit integers.
BRANCH=none
BUG=chromium:794010
TEST=Set CONFIG_LLSR_TEST, build, update cr50, and run llsrtest
on the console.
Change-Id: Iae66c86720c531454ba29f15b3cc6a07959f5ef2
Signed-off-by: Allen Webb <allenwebb@google.com>
Reviewed-on: https://chromium-review.googlesource.com/931932
Reviewed-by: Vadim Bendebury <vbendeb@chromium.org>
The ARMv7-M ISA defines standard (and optional) mechanism to manage the
CPU caches through the SCB (System Control Block) registers.
So far, only the Cortex-M7 core implements such as a mechanism (e.g. the
Cortex-M4 with caches we have are using a proprietary mechanism for the
management).
Define the functions to use the I-Cache,
and enable them on STM32H7 which is our only supported Cortex-M7 core.
The D-Cache mechanism is still To Be Done, as it involves a bit more
support in the firmware for the DMA memory areas.
Signed-off-by: Vincent Palatin <vpalatin@chromium.org>
BRANCH=none
BUG=b:67081508
TEST=on ZerbleBarn, verify manually that the 'IC' bit is set in the CCR
(e.g. 'rw 0xe000ed14' returns 0x60218), and runs some CPU workload
without crash and with a speed-up.
Change-Id: I6af1021d65048b787630387f7d95797db15d069c
Reviewed-on: https://chromium-review.googlesource.com/943445
Commit-Ready: Vincent Palatin <vpalatin@chromium.org>
Tested-by: Vincent Palatin <vpalatin@chromium.org>
Reviewed-by: Randall Spangler <rspangler@chromium.org>
The STM32H7 family has 2 banks of flash (with 2 hardware controllers
able to do 2 parallel operations at the same time).
Each bank of flash has 4 or 8 128-kB erase blocks (1MB and 2MB
variants).
The flash can only be written by 256-bit word (with an additional 10-bit
ECC computed by the hardware).
For the flash write-protection, we cannot use our 'classical' PSTATE
scheme as the erase-blocks are too large (128-kB) to dedicate one to
this and the embedded word in the RO partition would not work as the
flash has ECC and triggers bus-fault when the ECC is incorrect (which
includes the case where the 256-bit word is written a second time).
So we will do the following:
- use the RSS1 bit in the option bytes as the Write-Protect enabled bit.
- if the WP GPIO is set, lock at startup the option bytes until next
reboot.
Signed-off-by: Vincent Palatin <vpalatin@chromium.org>
BRANCH=none
BUG=b:67081508
TEST=run flashinfo/flashwp/flashwrite/flasherase commands on the EC
console.
Change-Id: I823fce3bd42b4df212cf0b8ceceaca84109b78e6
Reviewed-on: https://chromium-review.googlesource.com/901423
Commit-Ready: Vincent Palatin <vpalatin@chromium.org>
Tested-by: Vincent Palatin <vpalatin@chromium.org>
Reviewed-by: Nicolas Boichat <drinkcat@chromium.org>
In particular, this will allow touchpad driver and keyboard matrix
scanning to be powered off/disabled when the USB interface is
disabled without setting the remote wake feature
(USB_REQ_FEATURE_DEVICE_REMOTE_WAKEUP), as events would be
ignored anyway.
BRANCH=none
BUG=b:72683995
TEST=With next CLs, touchpad and keyboard matrix scanning are disabled
when lid is closed.
Change-Id: I3750bfaf8c31cde075adf9da4fef39753b8981c5
Signed-off-by: Nicolas Boichat <drinkcat@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/897067
Reviewed-by: Vincent Palatin <vpalatin@chromium.org>
stm32f0 has 20 bytes (not 20 words) of VBAT-backed RAM. Make more
efficient use of our limited storage to prevent trying to use storage
that doesn't exist.
BUG=b:71333840
BRANCH=None
TEST=Negotiate PD, run "reboot" on scarlet EC console, verify reset path
is taken in pd_partner_port_reset().
Signed-off-by: Shawn Nematbakhsh <shawnn@chromium.org>
Change-Id: Ie4c303b74a1b82b84ec971cdcc19c2b21a0032e7
Reviewed-on: https://chromium-review.googlesource.com/885461
Commit-Ready: Shawn N <shawnn@chromium.org>
Tested-by: Shawn N <shawnn@chromium.org>
Reviewed-by: Vincent Palatin <vpalatin@chromium.org>
The compiler marks data put into the TPM2_common.bss section as
PROGBITS, which the linker does not like. Changing the section name
prevents the marking and keeps linker happy.
BRANCH=cr50
BUG=chromium:799385
TEST=verified that local_state is still in where it belongs:
$ egrep '(local_state|__bss_libtpm2)' build/cr50/RW/ec.RW.smap
00010400 B __bss_libtpm2_start
00015d0c b local_state
00015d18 B __bss_libtpm2_end
Change-Id: I48f7d2cb08c7ccb2ef3b3159eaf4d66e2b8720b4
Signed-off-by: Vadim Bendebury <vbendeb@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/852793
Reviewed-by: Randall Spangler <rspangler@chromium.org>
For debug, in common code let's log the watchdog PC
and task id as our SW panic params.
BUG=chromium:790006
BRANCH=none
TEST=manually test scarlet rev2 from a1-a3, b1-b2:
(a1) Add 'while(1);' in button ISR
(a2) Boot and press the button
(a3) When watchdog is triggeried, check with 'panicinfo'
that saved R5 is the PC for button ISR.
(b1) 'crash watchdog' in EC console
(b2) Check with 'panicinfo' that CONSOLE task id is saved in
EXCEPTION and PC is saved in R5.
Change-Id: I64d2fcf594dd24b0951e002ab8e80ebcac2d1def
Signed-off-by: Philip Chen <philipchen@google.com>
Reviewed-on: https://chromium-review.googlesource.com/803618
Commit-Ready: Philip Chen <philipchen@chromium.org>
Tested-by: Philip Chen <philipchen@chromium.org>
Reviewed-by: Vincent Palatin <vpalatin@chromium.org>
- We have more flash space to use with nds32 toolchain GCC6.3.0,
so we enable a few console commands that were disabled previously.
And we also enable LTO to reduce the size of FW image.
- Put "__wait_evt" function into ram_code section to
fill the gap of flash and improving performance of code-fetch.
BUG=none
BRANCH=none
TEST=boot to kernel on reef_it8320.
Change-Id: I3b745ff80a57ef1163794864c39c22f7e1f86634
Signed-off-by: Dino Li <Dino.Li@ite.com.tw>
Reviewed-on: https://chromium-review.googlesource.com/788712
Commit-Ready: ChromeOS CL Exonerator Bot <chromiumos-cl-exonerator@appspot.gserviceaccount.com>
Reviewed-by: Randall Spangler <rspangler@chromium.org>
Add espi control module for it83xx.
Signed-off-by: Dino Li <dino.li@ite.com.tw>
BRANCH=none
BUG=none
TEST=1. it8390+Intel SKL-Y RVP3 and boot to shell.
2. console command "kbpress 1 4" to test keyboard data.
(board code for espi module test on CL:392587)
Change-Id: I1b32bd16f7e01abf07b9c9a68ebef2399cc9828d
Reviewed-on: https://chromium-review.googlesource.com/394471
Commit-Ready: Dino Li <Dino.Li@ite.com.tw>
Tested-by: Dino Li <Dino.Li@ite.com.tw>
Reviewed-by: Vincent Palatin <vpalatin@chromium.org>
Support protection of regions that aren't aligned to a power of 2 by
using two MPU entries, and taking advantage of the sub-region feature.
Also protect code RAM from being overwritten, on parts that use external
storage.
BUG=chromium:782244
BRANCH=None
TEST=On kevin, call:
mpu_protect_data_ram();
mpu_protect_code_ram();
mpu_enable();
Verify that first call results in the following update_region params:
addr: 0x200c2000 size: 0xc01d
Decoded: Protect 24K region
Verify that second call results in the following params:
addr: 0x100a8000 size: 0xc021
Decoded: Protect 96K region
addr: 0x100c0000 size: 0xf01b
Decoded: Protect remaining 8K region
Also verify that writes to beginning and end of code ram region trigger
data access violation after enabling protection.
Also verify that sysjump fails.
Change-Id: Ieb7a4ec3a089e8a2d29f231e1e3acf2e78e560a1
Signed-off-by: Shawn Nematbakhsh <shawnn@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/757721
Commit-Ready: Shawn N <shawnn@chromium.org>
Tested-by: Shawn N <shawnn@chromium.org>
Reviewed-by: Vincent Palatin <vpalatin@chromium.org>
With this change, we can pass "make buildall" at local
after it83xx based boards were removed from skip_boards.
BRANCH=none
BUG=none
TEST=- Passed "make buildall -j"
- CROSS_COMPILE_nds32=nds32le-cros-elf- make BOARD=it83xx_evb -j,
build ec image by using nds32le-cros-elf toolchain.
- make BOARD=it83xx_evb -j, coreboot-sdk is used.
Change-Id: I689b67ed50ac5c80e7526f157ba28733d7216e14
Signed-off-by: Dino Li <Dino.Li@ite.com.tw>
Reviewed-on: https://chromium-review.googlesource.com/762807
Commit-Ready: Patrick Georgi <pgeorgi@chromium.org>
Tested-by: Patrick Georgi <pgeorgi@chromium.org>
Reviewed-by: Patrick Georgi <pgeorgi@chromium.org>
The original assembly code of deep sleep bypass will cause build error
if both CONFIG_LOW_POWER_IDLE and CONFIG_LTO are defined when buildiing
board glkrvp/zoombini. This CL fixed it by change the bypass assembly code
from:
asm ("push {r0-r5}\n"
"ldr r0, =0x100A8000\n"
"wfi\n"
"ldm r0, {r0-r5}\n"
"pop {r0-r5}\n"
"isb\n"
);
to:
asm ("push {r0-r5}\n"
"wfi\n"
"ldm %0, {r0-r5}\n"
"pop {r0-r5}\n"
"isb\n" :: "r" (0x100A8000)
);
BRANCH=none
BUG=none
TEST=No build errors for "make buildall".
TEST=build zoombini/glkrvp with CONFIG_LOW_POWER_IDLE and CONFIG_LTO,
no build errors.
TEST=build npcx7_evb/npcx_evb and do stress test for deep idle->wakeup
on EVB, no symptom observed.
Change-Id: I90b13b4baf418e3f4b3234d4811e3978b6436aac
Signed-off-by: Vincent Palatin <vpalatin@chromium.org>
Signed-off-by: CHLin <CHLIN56@nuvoton.com>
Reviewed-on: https://chromium-review.googlesource.com/756535
Commit-Ready: CH Lin <chlin56@nuvoton.com>
Tested-by: CH Lin <chlin56@nuvoton.com>
old_val may change value after cmpxchg, need to reset to 0 for next loop
comparison.
BUG=None
BRANCH=master
TEST=On Soraka modified for ISH board, running more than 4 hours sensor
data fetching by host command, and no problem.
Change-Id: I720230e196771071c8ba204458da6c4788d374ea
Signed-off-by: li feng <li1.feng@intel.com>
Reviewed-on: https://chromium-review.googlesource.com/738914
Commit-Ready: Li1 Feng <li1.feng@intel.com>
Tested-by: Li1 Feng <li1.feng@intel.com>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
A built-in function (__builtin_ffs) is missing after we sync down
the source code to latest.
Signed-off-by: Dino Li <dino.li@ite.com.tw>
BRANCH=none
BUG=none
TEST=We use the CL:730246 to verify this change, and
no error was received after running the test.
Change-Id: I5210f85db05650545c9924940e8b24e350b82f71
Reviewed-on: https://chromium-review.googlesource.com/730245
Commit-Ready: Dino Li <Dino.Li@ite.com.tw>
Tested-by: Dino Li <Dino.Li@ite.com.tw>
Reviewed-by: Vincent Palatin <vpalatin@chromium.org>
NPCX5* only has one UART controller, which can be switched
between 2 pads. We keep the default pad for EC console,
however, we allow switching to the alternate pad for short,
infrequent, transactions. Both pads are assumed to use the
same baudrate and other line settings.
When switching pad, we first configure the new pad, then switch
off the old one, to avoid having no pad selected at a given time,
see b/65526215#c26.
Because of the added complexity of npcx_gpio2uart (and the fact
that it uses the global variable "pad" define in uart.c), we
move the implementation to uart.c (npcx_uart2gpio is also moved
for consistency).
When the pad is switched to alternate pad, characters input
and output on the EC console (default pad) would be lost. To
compensate for this, we:
- Switch back to main pad in case of EC panic, so that output
is shown on EC console.
- Immediately abort current alternate pad transaction if a
character is received on the default pad. Note, however,
that the first character will be lost (this can be worked
around by telling user to press enter, and have servod/FAFT
always send 2 blank lines (instead of just one) before
sending a command).
- Inhibit pad switching for 500ms after receiving a character
on default pad. Assuming a reasonable typing speed, this
should allow developers to type console commands relatively
comfortably, while not starving the alternate pad communication
for too long.
The logic above could be simplified significantly by implementing
software flow control (XON/XOFF, see b/67026316).
BRANCH=none
BUG=b:65526215
TEST=While follow-up CL that writes long 1k buffers, the following
works fine:
- type 'uart' in EC console
- Read battery power consumption from servod, which "types" in
the EC console:
while true; do dut-control ppvar_vbat_mw; sleep 1; done
no failure is seen.
TEST=Add this test code in uart_alt_pad_read_write, after the pad
has been switched, and check that panic information is
consistently printed correctly:
{
static int t;
if (t++ > 20)
t = t / ret;
}
Change-Id: I18feed2f8ca4eb85f40389f77dac3a46315310e7
Signed-off-by: Nicolas Boichat <drinkcat@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/659458
Reviewed-by: Vincent Palatin <vpalatin@chromium.org>
Modules in board/cr50/tpm2 are part of the TPM task, and their .bss
segment needs to be wiped out along with other TPM modules' when TPM
is reset.
The only module from this directory which has static variables is
NVMem.c. This patch groups the variables into a structure, and makes
sure that the structure is stored in the part of .bss which is zeroed
when TPM is reset.
BRANCH=none
BUG=none
TEST=verified that TPM wipeout happening during RMA reset is
successful.
Change-Id: I5f80dc5e4f5f7cdce2d1a1ed59e0fbac354c5bf6
Signed-off-by: Vadim Bendebury <vbendeb@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/733814
Reviewed-by: Randall Spangler <rspangler@chromium.org>
Reviewed-by: Scott Collyer <scollyer@chromium.org>
In order to prevent keeping the CSAE bit at 1 forever impacts the eSPI
performance, the npcx driver enables host access wakeup functionality
before ec enters deep sleep or wfi. But this bypass also should be added
in __idle() of core/cortex-m/task.c if CONFIG_LOW_POWER_IDLE is
disabled.
This CL also narrows the bypass only when host interface is eSPI.
BRANCH=eve
BUG=b:64730183
TEST=No build errors for make buildall. Disable CONFIG_LOW_POWER_IDLE
functionality on poppy and use following script "count=0; while :;
do echo "--- iteration --- $count"; time flashrom -p ec -r ec.bin; sleep
1; count=$((${count}+1)); done" to test eSPI performances over 300
times. No errors occur and all tests' efficiency are the same as
removing CSAE bypass.
Change-Id: I8b6b69e37318208c185747151c06b3e6bdfd2f4e
Signed-off-by: Mulin Chao <mlchao@nuvoton.com>
Reviewed-on: https://chromium-review.googlesource.com/644967
Reviewed-by: Daisuke Nojiri <dnojiri@chromium.org>
CCD_CHANGE_HOOK should use CONFIG_CASE_CLOSED_DEBUG_V1.
All boards which use chip/g either use both CONFIG_USB_SERIALNO and
CONFIG_CASE_CLOSED_DEBUG or neither of them, so just depend on
CONFIG_USB_SERIALNO.
This is in preparation for making common/case_closed_debug refer only
to the usb_pd_protocol version (with mode=disabled/partial/enabled),
and cr50 will have its own version (with only enabled/disabled, and
tied more closely to CCD config).
No functionality changes.
BUG=none
BRANCH=cr50
TEST=make buildall -j; boot cr50 and see change hook called
Change-Id: I1985c8c48c1a85fed4549402a7b47b8a9cf135d7
Signed-off-by: Randall Spangler <rspangler@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/648067
Reviewed-by: Mary Ruthven <mruthven@chromium.org>
The way Cortex processors handle exceptions allows writing exception
routines directly in C, as return from exception is handled by providing
a special value for the link register.
However, it is not safe to do this when doing context switching. In
particular, C handlers may push some general-purpose registers that
are used by the handler and pop them later, even when context switch
has happened in the meantime. While the processor will restore {r0-r3}
from the stack when returning from an exception, the C handler code
may push, use and pop another register, such as r4.
It turns out that GCC 4.8 would generally only use r3 in svc_handler and
pendsv_handler, but newer versions tend to use r4, thus clobbering r4
that was restored from the context switch and leading up to a fault
when r4 is used by the task code.
An occurrence of this behaviour takes place with GCC > 4.8 in __wait_evt,
where "me" is stored in r4, which gets clobbered after an exception
triggers pendsv_handler. The exception handler uses r4 internally, does
a context switch and then restores the previous value of r4, which is
not restored by the processor's internal, thus clobbering r4.
This ends up with the following assertion failure:
'tskid < TASK_ID_COUNT' in timer_cancel() at common/timer.c:137
For this reason, it is safer to have assembly routines for exception
handlers that do context switching.
BUG=chromium:631514
BRANCH=None
TEST=Build and run speedy EC with a recent GCC version
Change-Id: Ib068bc12ce2204aee3e0f563efcb94f15aa87013
Signed-off-by: Paul Kocialkowski <contact@paulk.fr>
Reviewed-on: https://chromium-review.googlesource.com/362830
Commit-Ready: Shawn N <shawnn@chromium.org>
Tested-by: Shawn N <shawnn@chromium.org>
Reviewed-by: Shawn N <shawnn@chromium.org>
For system using core/minute-ia, build will fail since __image_size is
not defined. If later this variable is used, should adjust its value
properly.
BUG=none
BRANCH=None
TEST=Build passed for Soarka ISH which uses core/minute-ia.
Change-Id: I8e179e0bac551a46d93ca10ba8b61b4ebade74fc
Signed-off-by: li feng <li1.feng@intel.com>
Reviewed-on: https://chromium-review.googlesource.com/595151
Commit-Ready: Li1 Feng <li1.feng@intel.com>
Tested-by: Li1 Feng <li1.feng@intel.com>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
This patch defines CONFIG_DATA_RAM_SIZE, which indicates the size
of the RAM used for data, thus can be marked as non-executable.
If it's not defined, it defaults to CONFIG_RAM_SIZE. Thus, other chips
are not affected.
BUG=b:36037354
BRANCH=none
TEST=buildall. Run 'sysjump disable' on Reef and verify mpu_protect_ram
is successful.
Change-Id: I54d74fd1dabff7e1013fff2542fd02c3646803d1
Signed-off-by: Daisuke Nojiri <dnojiri@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/596518
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
This adds the CCD configuration module, and the console commands to
control it. It is not wired up to any of the CCD capabilities; that's
coming in the next CL.
Briefly:
* CCD configuration is persistently stored in nvmem_vars. Use ccdinfo to
print it.
* CCD can be Locked, Unlocked (some capabilities), or Opened
(all capabilities), using the ccdlock / ccdunlock / ccdopen commands.
* CCD config can be restricted by setting a password via ccdpass.
* Individual config capabilities can be set via ccdset. Some of those will
be used to gate access to things like write protect and UARTs. Others
affect the requirements for ccdunlock / ccdopen (for example, is physical
presenc required).
* The entire config can be reset via ccdreset. If only unlocked, config
that is restricted to Opened is not reset.
* If CR50_DEV=1, ccdoops will force-reset and open the config.
See go/cr50-ccd-wp for more information.
BUG=b:62537474
BRANCH=none
TEST=manual with CR50_DEV=1 build
gpioget # make sure GPIO_BATT_PRES_L=0
ccdlock # lock, because CR50_DEV=1 builds start unlocked
ccdinfo # locked, flags=0, all capabilities default
ccdpass # access denied (we're locked)
ccdreset # access denied
ccdset flashap always # access denied
ccdunlock
ccdinfo # unlocked
ccdpass foo
ccdinfo # flags=2 (password set when unlocked)
ccdset flashap always # access denied
ccdset uartectx unlesslocked
ccdinfo # yes, uartectx permission changed
ccdlock
ccdunlock # fails without password
ccdunlock bar # wrong password
ccdunlock foo # busy
(wait 3 sec)
ccdunlock foo
ccdreset
ccdinfo # no password, flags 0, capabilities all default
ccdopen # requires physical presence; tap power or use 'pp'
ccdset uartectx unlesslocked
ccdset batterybypasspp ifopened
ccdpass baz
ccdinfo # password set, flag 0, ccdset changes worked
ccdunlock
ccdreset
ccdinfo # uartectx back to ifopened, password still set
ccdopen baz # still requires physical presence
ccdset opennolongpp always
ccdlock
ccdopen baz # no pp required
ccdset unlocknoshortpp unlesslocked
ccdlock
ccdopen baz # short pp sequence required (3 taps)
ccdlock
ccdunlock baz # short pp sequence required
ccdopen baz # pp not required
ccdset unlocknoshortpp always
ccdlock
testlab open # access denied
testlab enable # access denied
ccdunlock baz
testlab open # access denied
testlab enable # access denied
ccdopen baz
testlab enable # requires short pp
ccdinfo # flags 1
ccdreset
ccdinfo # no password, flags=1, caps all default
ccdlock
testlab open
ccdinfo # opened
testlab disable # requires short pp; let it time out
ccdinfo # still opened, flags=1
ccdlock
ccdoops # backdoor in CR50_DEV images to force-reset CCD
ccdinfo # opened, flags=0, all defaults (yes, oops wipes out testlab)
ccdreset rma
ccdinfo # flags = 0x400000, everything but Cr50FullConsole always
ccdreset # back to flags=0, all default
Change-Id: I24e8d8f361874671e6e94f27492ae00db919bea9
Reviewed-on: https://chromium-review.googlesource.com/569439
Commit-Ready: Randall Spangler <rspangler@chromium.org>
Tested-by: Randall Spangler <rspangler@chromium.org>
Reviewed-by: Vadim Bendebury <vbendeb@chromium.org>
Instead of having to pass in both the board and the arch used there
(which might be complicated in the light of different ECs with different
arches), allow passing in the board and a set of cross compilers.
The core/*/build.mk then pick the compiler that is responsible for them.
The current method works just the same: If you've already set
CROSS_COMPILE, no override happens. If you set neither CROSS_COMPILE nor
CROSS_COMPILE_$arch, the same default as before this CL is set.
BUG=none
BRANCH=none
TEST=emerge-{samus,kevin} chromeos-ec behave reasonably with adapted
ebuild
Change-Id: Icf1866f296412dd92ecfe134394224c49f7c3df5
Signed-off-by: Patrick Georgi <pgeorgi@google.com>
Reviewed-on: https://chromium-review.googlesource.com/549344
Commit-Ready: Patrick Georgi <pgeorgi@chromium.org>
Tested-by: Patrick Georgi <pgeorgi@chromium.org>
Reviewed-by: Stefan Reinauer <reinauer@google.com>
Remove code related to DH_SWAP_BY_POINTERS and
DH_REPLACE_LAST_THREE_LADDERSTEPS_WITH_DOUBLINGS, as they are
not really worth it (code size increase of 60/100 bytes respectively
for 1000/500 us time gain).
BRANCH=none
BUG=b:62813194
TEST=make BOARD=hammer PROJECT=x25519 TEST_BUILD=y
./util/flash_ec --board=hammer --image=build/hammer/x25519.bin
EC console: runtest, taskinfo
Change-Id: If21948bb2c7c20f97ba8b321dd2688cd3d0ba74a
Reviewed-on: https://chromium-review.googlesource.com/554440
Commit-Ready: Nicolas Boichat <drinkcat@chromium.org>
Tested-by: Nicolas Boichat <drinkcat@chromium.org>
Reviewed-by: Vincent Palatin <vpalatin@chromium.org>
- Move generic implementation to curve25519-generic.o
- Always use optimized version on cortex-m0.
- Rename .s files to .S, remove unnecessary lines in assembly files.
- Rename crypto_scalarmult_curve25519 to x25519_scalar_mult to match
the signature provided by the generic implementation.
- Replace some handcoded memcpy with function calls
- Remove unnecessary "volatile" specifications in the code.
BRANCH=none
BUG=b:62813194
TEST=To test old implementation only:
- Increase CONFIG_RO_SIZE to 60kb
- Increase console stack size to 2048
make BOARD=hammer PROJECT=x25519 TEST_BUILD=y
./util/flash_ec --board=hammer --image=build/hammer/x25519.bin
EC console: runtest, taskinfo
=> Used to takes ~4'17" to run (X25519 duration 256347 us).
1496/2048 stack size usage in CONSOLE task
=> Now takes ~1'25" to run (X25519 duration 84520 us)
732/2048 stack size usage in CONSOLE task
TEST=In test/x25519.c, uncomment #define TEST_X25519_1M_ITERATIONS
make BOARD=hammer PROJECT=x25519 TEST_BUILD=y
./util/flash_ec --board=hammer --image=build/hammer/x25519.bin
EC console: runtest, wait ~23 hours, test passes.
TEST=- Define CONFIG_CURVE25519_CORTEXM0 (next patch)
makes newsizes
build/hammer/RW/ec.RW.flat shrank by 1888 bytes: (52208 to 50320)
Change-Id: Icce38d3c32f431a85ac0f951cf34456b490dc665
Reviewed-on: https://chromium-review.googlesource.com/540962
Commit-Ready: Nicolas Boichat <drinkcat@chromium.org>
Tested-by: Nicolas Boichat <drinkcat@chromium.org>
Reviewed-by: Nicolas Boichat <drinkcat@chromium.org>
We need to add a few library routines for sensor task.
Routines added are taken from nds32's library.
BRANCH=none
BUG=none
TEST=Add sensor task to reef_it8320 board.
Test screen rotation functionality on reef_it8320.
Change-Id: I2eee33f897b38e05bddd30b16f875944259b2c0d
Signed-off-by: Dino Li <Dino.Li@ite.com.tw>
Reviewed-on: https://chromium-review.googlesource.com/527537
Reviewed-by: Randall Spangler <rspangler@chromium.org>
Reviewed-by: Vincent Palatin <vpalatin@chromium.org>
Until HOOK_INIT has completed, do not allow any tasks other than HOOKS
or IDLE to be scheduled. Programmers often make the assumption that
a HOOK_INIT function is guaranteed to be run before task code that depends
on it, so let's make it so.
BUG=chromium:649398
BRANCH=None
TEST=Manual on kevin, compare boot without patch:
...
[0.004 power state 0 = G3, in 0x0008] <-- from chipset task
RTC: 0x00000000 (0.00 s)
[0.004 power state 4 = G3->S5, in 0x0008]
RTC: 0x00000000 (0.00 s)
[0.005 clear MKBP fifo]
[0.006 clear MKBP fifo]
[0.006 KB init state: ... <-- from keyscan task
[0.012 SW 0x05]
[0.155 hash start 0x00020000 0x00019a38]
[0.158 HOOK_INIT DONE!]
... to boot with patch:
...
RTC: 0x58cc614c (1489789260.00 s)
[0.004 clear MKBP fifo]
[0.005 clear MKBP fifo]
[0.010 SW 0x05]
[0.155 hash start 0x00020000 0x000198e0]
[0.157 HOOK_INIT DONE!]
...
Also, verify kevin boots to OS and is generally functional through
sysjump and basic tasks, and verify elm (stm32f0 / cortex-m0) boots.
Change-Id: If56fab05ce9b9650feb93c5cfc2d084aa281e622
Signed-off-by: Shawn Nematbakhsh <shawnn@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/456628
Commit-Ready: Shawn N <shawnn@chromium.org>
Tested-by: Shawn N <shawnn@chromium.org>
Reviewed-by: Vincent Palatin <vpalatin@chromium.org>
Cortex-M4 supports a floating point square root function that takes 14
cycles to execute, which is a speed improvement over the existing binary
search, and saves flash space.
BUG=chromium:687624
BRANCH=None
TEST=On kevin, verify that both sqrtf methods (binary search vs HW
instruction) have identical results for fractional input (eg.
sqrt(15.999999) = 3), except when floating point representation of
square root rounds up to an integer. Verify identical results for all
integers [-100, 16793602). Note that 16793602 is the first integer for
which the floating point representation of sqrt rounds up to an integer.
Also verify basic motion sense functions on kevin.
Signed-off-by: Shawn Nematbakhsh <shawnn@chromium.org>
Change-Id: I8521c9a28c958b340ca83c37342253e424df0c91
Reviewed-on: https://chromium-review.googlesource.com/537734
Commit-Ready: Shawn N <shawnn@chromium.org>
Tested-by: Shawn N <shawnn@chromium.org>
Reviewed-by: Shawn N <shawnn@chromium.org>
Because N8 CPU doesn't have floating point unit,
so we implement an extra floating point engine
(single-precision addition, subtraction, multiplication,
and division) into it8320 to improve performance of
floating point operation.
To make CPU's instruction compatible, we use register (DLMB)
to switch ALU (Arithmetic Logic Unit). eg:
Instruction 'ADD45' adds the contents of two registers then
writes the result to the source register.
But if we switch ALU to floating point operation mode,
this instruction will do a floating-point addition instead.
For the other FPU that we don't support as far,
we have to use soft float library routines of nds32.
Signed-off-by: Dino Li <dino.li@ite.com.tw>
BRANCH=none
BUG=none
TEST=add the following console command and test different
scenarios by changing variable a and b.
#define PRINTF_FLOAT(x) ((int)((x) * 1000.0f))
static int it83xx_fpu_test(int argc, char **argv)
{
volatile float a = 1.23f;
volatile float b = 4.56f;
volatile float c;
c = a + b;
ccprintf("__addsf3: (%d)\n", PRINTF_FLOAT(c));
c = a - b;
ccprintf("__subsf3: (%d)\n", PRINTF_FLOAT(c));
c = a * b;
ccprintf("__mulsf3: (%d)\n", PRINTF_FLOAT(c));
c = a / b;
ccprintf("__divsf3: (%d)\n", PRINTF_FLOAT(c));
return EC_SUCCESS;
}
DECLARE_CONSOLE_COMMAND(fpu, it83xx_fpu_test, "", "");
Change-Id: I4fc1c08d8c2376156bec9f098491187675c4a88f
Reviewed-on: https://chromium-review.googlesource.com/427640
Commit-Ready: Dino Li <Dino.Li@ite.com.tw>
Tested-by: Dino Li <Dino.Li@ite.com.tw>
Reviewed-by: Vincent Palatin <vpalatin@chromium.org>
This CL implements two methods for hibernating on npcx7 ec. One is using
PSL (Power Switch Logic) circuit to cut off ec's VCC power rail. The
other is turning off the power of all ram blocks except the last code
ram block. In order to make sure hibernate utilities are located in the
last code ram block and work properly, we introduce a new section called
'after_init' in ec.lds.S.
We also moved the hibernate utilities, workarounds for sysjump and so on
which are related to chip family into system-npcx5/7.c. It should be
easier to maintain.
It also includes:
1. Add CONFIG_HIBERNATE_PSL to select which method is used on npcx7 for
hibernating.
2. Add new flag GPIO_HIB_WAKE_HIGH to configure the active priority of
wake-up inputs during hibernating.
3. Add DEVICE_ID for npcx796f.
BRANCH=none
BUG=none
TEST=No build errors for all boards using npcx5 series.
Build poppy board and upload FW to platform. No issues found. Make
sure AC_PRESENT and POWER_BUTTON_L can wake up system from
hibernate. Passed hibernate tests no matter CONFIG_HIBERNATE_PSL is
enabled or not on npcx796f evb.
Change-Id: I4e045ebce4120b6fabaa582ed2ec31b5335dfdc3
Signed-off-by: Mulin Chao <mlchao@nuvoton.com>
Reviewed-on: https://chromium-review.googlesource.com/493006
Reviewed-by: Randall Spangler <rspangler@chromium.org>
Image used size is now part of the image_data struct present in all
images at a fixed offset, so use it rather than scanning from the end of
the image.
BUG=chromium:577915
TEST=Verify on kevin + lars + lars_pd that system_get_image_used() returns
the same value as the old implementation, for both RO and RW images.
BRANCH=None
Change-Id: I35f0aa87f5ab1371dbd8b132f22b9d0044358223
Signed-off-by: Shawn Nematbakhsh <shawnn@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/450859
Commit-Ready: Shawn N <shawnn@chromium.org>
Tested-by: Shawn N <shawnn@chromium.org>
Reviewed-by: Randall Spangler <rspangler@chromium.org>
Store our image size (known at build time) in our version struct (now
renamed to image_data). This will allow us to more efficiently determine
the size of an image in a follow-up CL.
Note that compatibility is broken for old ROs that do not include this
CL.
BUG=chromium:577915
TEST=Verify on kevin + lars + lars_pd that stored image size matches
output of system_get_image_used() for both RO and RW images.
BRANCH=None
Change-Id: I7b8dc3ac8cf2df3184d0701a0e0ec8032de8d81b
Signed-off-by: Shawn Nematbakhsh <shawnn@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/450858
Commit-Ready: Shawn N <shawnn@chromium.org>
Tested-by: Shawn N <shawnn@chromium.org>
Reviewed-by: Randall Spangler <rspangler@chromium.org>
This marks decriptor length as byte accurate, by
aligning the "__usb_desc_end" label by byte rather
than short or word.
BUG=chrome-os-partner:62873
TEST=servo_v4 doesn't crash adb anymore
BRANCH=None
Change-Id: I643bae2c59e81ce0c03d9026f4a5d7933ba4c891
Reviewed-on: https://chromium-review.googlesource.com/442012
Commit-Ready: Nick Sanders <nsanders@chromium.org>
Tested-by: Nick Sanders <nsanders@chromium.org>
Reviewed-by: Vincent Palatin <vpalatin@chromium.org>
