drivers: STM32 dualcore concurrent register access protection with HSEM #24862

ABOSTM · 2020-04-30T16:39:23Z

In case of dualcore, STM32H7, STM32W and STM32MP1,
protect concurrent register write access with HSEM.
Done for following drivers:
clock_control, counter, flash, gpio, interrupt_controller

Signed-off-by: Alexandre Bourdiol [email protected]

zephyrbot · 2020-04-30T16:42:00Z

All checks are passing now.

Tip: The bot edits this comment instead of posting a new one, so you can check the comment's history to see earlier messages.

drivers/gpio/gpio_stm32.c

drivers/clock_control/clock_stm32_ll_h7.c

drivers/counter/counter_ll_stm32_rtc.c

soc/arm/st_stm32/common/stm32_hsem.h

drivers/clock_control/CMakeLists.txt

soc/arm/st_stm32/stm32h7/soc.h

drivers/clock_control/clock_stm32_ll_h7.c

soc/arm/st_stm32/stm32mp1/soc.h

soc/arm/st_stm32/stm32wb/soc.h

soc/arm/st_stm32/common/stm32_hsem.h

pabigot

Multiple functions have an exit path while the lock is still held. These should be fixed.

Also it appears that there are nested lock calls; I don't know the hardware to say whether this is safe.

drivers/clock_control/clock_stm32_ll_h7.c

drivers/clock_control/clock_stm32_ll_mp1.c

soc/arm/st_stm32/common/stm32_hsem.h

pabigot

One clear problem newly identified, and an overall impression that the solution needs to be reconsidered to take into account protecting the lock status from mutation by interrupts. The latter is outside my area of expertise.

drivers/clock_control/clock_stm32_ll_h7.c

drivers/flash/flash_stm32.c

ABOSTM · 2020-06-08T09:37:47Z

I made a rebase of this PR.
Is there further comments?
or can it be reviewed and merged ?

Obsolete

pabigot

Just commentary: I don't have expertise to judge the technical merit of this SOC-specific solution.

I've removed my NAK under the assumption that the mis-ordered lock sequences have been fixed, and each lock is paired with an unlock (no early exits/inappropriate nested locking). From a cursory look only at the diffs github provides this seems to be true, but I have not looked at the unmodified areas of the code to confirm it.

I find it confusing that z_stm32_hsem_lock is paired with z_stm32_hsem_release. Pairing lock with unlock, or request with release, would make be more consistent with existing usage (request/release only if the lock is recursive, but I think it isn't). Since this is STM32-specific API I don't insist on this.

soc/arm/st_stm32/stm32mp1/soc.h

ABOSTM · 2020-06-23T16:38:08Z

I replaced z_stm32_hsem_release() by z_stm32_hsem_unlock().
I also fixed rebase regression (include model).

ABOSTM · 2020-06-24T07:02:18Z

Sanitychek failure: ' Sanitycheck / qemu_arc_hs:tests/portability/cmsis_rtos_v1/portability.cmsis_rtos_v1 / tests/portability/cmsis_rtos_v1/portability.cmsis_rtos_v1'

arnopo

just a minor comment, else LGTM

drivers/clock_control/clock_stm32_ll_h7.c

In case of dualcore, STM32H7, STM32W and STM32MP1, protect concurrent register write access with HSEM. Done for following drivers: clock_control, counter, flash, gpio, interrupt_controller Signed-off-by: Alexandre Bourdiol <[email protected]>

Due to HSEM implementation zephyrproject-rtos#24862, USB CLK48 lock implementation zephyrproject-rtos#25850 should be reworked. And by the way, implement the same in entropy which is using the same clock. Signed-off-by: Alexandre Bourdiol <[email protected]>

Due to HSEM implementation #24862, USB CLK48 lock implementation #25850 should be reworked. And by the way, implement the same in entropy which is using the same clock. Signed-off-by: Alexandre Bourdiol <[email protected]>

ABOSTM requested review from andrewboie, arnopo, erwango, nordic-krch and superna9999 as code owners April 30, 2020 16:39

zephyrbot added the area: Counter label Apr 30, 2020

erwango requested changes Apr 30, 2020

View reviewed changes

drivers/gpio/gpio_stm32.c Outdated Show resolved Hide resolved

drivers/clock_control/clock_stm32_ll_h7.c Outdated Show resolved Hide resolved

drivers/clock_control/clock_stm32_ll_h7.c Outdated Show resolved Hide resolved

erwango added platform: STM32 ST Micro STM32 and removed area: Counter labels Apr 30, 2020

gmarull reviewed Apr 30, 2020

View reviewed changes

erwango mentioned this pull request May 4, 2020

Ethernet support for STM32H747 #24767

Closed

ABOSTM force-pushed the STM32_DUALCORE_HSEM branch from 3bb5ece to 02e63af Compare May 4, 2020 14:22

ABOSTM requested review from mnkp, nashif, nvlsianpu and pabigot as code owners May 4, 2020 14:22

zephyrbot added the area: Counter label May 4, 2020

gmarull reviewed May 4, 2020

View reviewed changes

soc/arm/st_stm32/common/stm32_hsem.h Outdated Show resolved Hide resolved

ABOSTM force-pushed the STM32_DUALCORE_HSEM branch from 02e63af to a45e947 Compare May 4, 2020 14:58

erwango reviewed May 4, 2020

View reviewed changes

drivers/clock_control/CMakeLists.txt Outdated Show resolved Hide resolved

erwango requested changes May 4, 2020

View reviewed changes

ABOSTM force-pushed the STM32_DUALCORE_HSEM branch from a45e947 to b7431a6 Compare May 4, 2020 17:12

pabigot requested changes May 4, 2020

View reviewed changes

drivers/clock_control/clock_stm32_ll_h7.c Outdated Show resolved Hide resolved

drivers/clock_control/clock_stm32_ll_h7.c Outdated Show resolved Hide resolved

ABOSTM force-pushed the STM32_DUALCORE_HSEM branch 2 times, most recently from a0c2f13 to 56f51c7 Compare May 5, 2020 11:46

arnopo requested changes May 5, 2020

View reviewed changes

drivers/clock_control/clock_stm32_ll_mp1.c Outdated Show resolved Hide resolved

soc/arm/st_stm32/common/stm32_hsem.h Outdated Show resolved Hide resolved

pabigot self-requested a review May 5, 2020 13:08

pabigot previously requested changes May 5, 2020

View reviewed changes

drivers/clock_control/clock_stm32_ll_h7.c Outdated Show resolved Hide resolved

drivers/flash/flash_stm32.c Show resolved Hide resolved

ABOSTM force-pushed the STM32_DUALCORE_HSEM branch from 56f51c7 to bb063fc Compare May 6, 2020 14:27

ABOSTM force-pushed the STM32_DUALCORE_HSEM branch from bb063fc to 6102ba1 Compare May 6, 2020 14:34

erwango added this to the v2.4.0 milestone May 25, 2020

ABOSTM force-pushed the STM32_DUALCORE_HSEM branch from 6102ba1 to 85b40a7 Compare June 5, 2020 16:56

ABOSTM force-pushed the STM32_DUALCORE_HSEM branch from 85b40a7 to ed06b8f Compare June 9, 2020 06:53

ABOSTM force-pushed the STM32_DUALCORE_HSEM branch 2 times, most recently from 631a94f to a928170 Compare June 23, 2020 12:35

erwango requested review from erwango, gmarull, pabigot and arnopo June 23, 2020 15:30

pabigot reviewed Jun 23, 2020

View reviewed changes

soc/arm/st_stm32/stm32mp1/soc.h Show resolved Hide resolved

ABOSTM force-pushed the STM32_DUALCORE_HSEM branch from a928170 to 50a581d Compare June 23, 2020 16:34

arnopo approved these changes Jul 7, 2020

View reviewed changes

drivers/clock_control/clock_stm32_ll_h7.c Outdated Show resolved Hide resolved

ABOSTM force-pushed the STM32_DUALCORE_HSEM branch from 50a581d to 6c6e445 Compare July 7, 2020 09:44

ABOSTM force-pushed the STM32_DUALCORE_HSEM branch from 6c6e445 to 8a51ebf Compare July 7, 2020 14:40

erwango approved these changes Jul 8, 2020

View reviewed changes

erwango requested a review from galak July 8, 2020 15:30

carlescufi merged commit c8ceca2 into zephyrproject-rtos:master Jul 9, 2020

KozhinovAlexander mentioned this pull request Aug 5, 2020

stm32h7 add entropy generator support #27374

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

drivers: STM32 dualcore concurrent register access protection with HSEM #24862

drivers: STM32 dualcore concurrent register access protection with HSEM #24862

ABOSTM commented Apr 30, 2020 •

edited

Loading

zephyrbot commented Apr 30, 2020 •

edited

Loading

pabigot left a comment

pabigot left a comment

ABOSTM commented Jun 8, 2020

pabigot left a comment •

edited

Loading

ABOSTM commented Jun 23, 2020

ABOSTM commented Jun 24, 2020

arnopo left a comment

drivers: STM32 dualcore concurrent register access protection with HSEM #24862

drivers: STM32 dualcore concurrent register access protection with HSEM #24862

Conversation

ABOSTM commented Apr 30, 2020 • edited Loading

zephyrbot commented Apr 30, 2020 • edited Loading

pabigot left a comment

Choose a reason for hiding this comment

pabigot left a comment

Choose a reason for hiding this comment

ABOSTM commented Jun 8, 2020

pabigot left a comment • edited Loading

Choose a reason for hiding this comment

ABOSTM commented Jun 23, 2020

ABOSTM commented Jun 24, 2020

arnopo left a comment

Choose a reason for hiding this comment

ABOSTM commented Apr 30, 2020 •

edited

Loading

zephyrbot commented Apr 30, 2020 •

edited

Loading

pabigot left a comment •

edited

Loading