The UEFI spec does not require interrupts to be disabled when invoking
runtime services. The reason we have been doing so is because EFI pstore
may call SetVariable() from interrupt context, which may result in deadlock
if another runtime services call was in progress on the same cpu.
The EFI pstore has already been updated to use a non-blocking path and fail
gracefully rather than spin forever, so we can updated the ordinary blocking
wrappers to run with interrupts enabled instead. This aims to prevent excessive
interrupt latencies on uniprocessor platforms with slow variable stores.
Changes since v2:
- added nonblocking QueryVariableInfo() wrapper variant, and updated
efi_query_variable_store() to use it when called in nonblocking context
- add patch to drop the rtc_lock spinlock
- add patch to drop redundant efi_set_variable_nonblocking_t typedef
- drop BUG_ONs in patch #7
Ard Biesheuvel (7):
efi: expose non-blocking set_variable() wrapper to efivars
efi: remove redundant efi_set_variable_nonblocking prototype
efi: runtime-wrappers: add a nonblocking version of QueryVariableInfo
efi: add nonblocking option to efi_query_variable_store()
efi: runtime-wrappers: remove out of date comment regarding in_nmi()
efi: runtime-wrapper: get rid of the rtc_lock spinlock
efi: runtime-wrappers: run UEFI Runtime Services with interrupts
enabled
arch/x86/platform/efi/quirks.c | 37 +++++--
drivers/firmware/efi/efi.c | 1 +
drivers/firmware/efi/runtime-wrappers.c | 115 +++++++-------------
drivers/firmware/efi/vars.c | 16 ++-
include/linux/efi.h | 21 ++--
5 files changed, 94 insertions(+), 96 deletions(-)
--
1.9.1
--
To unsubscribe from this list: send the line "unsubscribe linux-efi" in
the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
