Take an effort to recode the arm64 vdso code from assembler to C
previously submitted by Andrew Pinski <[email protected]>, rework
it for use in both arm and arm64, overlapping any optimizations
for each architecture. But instead of landing it in arm64, land the
result into lib/vdso and unify both implementations to simplify
future maintenance. This will act as the basis for implementing
arm64 vdso32 in the future.
[email protected] made the following claims in the original patch:
This allows the compiler to optimize the divide by 1000 and remove
the other divides.
On ThunderX, gettimeofday improves by 32%. On ThunderX 2,
gettimeofday improves by 18%.
Note I noticed a bug in the old implementation of __kernel_clock_getres;
it was checking only the lower 32bits of the pointer; this would work
for most cases but could fail in a few.
Signed-off-by: Mark Salyzyn <[email protected]>
Cc: James Morse <[email protected]>
Cc: Russell King <[email protected]>
Cc: Catalin Marinas <[email protected]>
Cc: Will Deacon <[email protected]>
Cc: Andy Lutomirski <[email protected]>
Cc: Dmitry Safonov <[email protected]>
Cc: John Stultz <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Laura Abbott <[email protected]>
Cc: Kees Cook <[email protected]>
Cc: Ard Biesheuvel <[email protected]>
Cc: Andy Gross <[email protected]>
Cc: Kevin Brodsky <[email protected]>
Cc: Andrew Pinski <[email protected]>
Cc: [email protected]
Cc: [email protected]
v2:
- split first CL into 7 pieces, there were cosmetic adjustments.
- make sure profiling is turned off.
- kept quiet_cmd_vdsoas.
v3:
- changed are a result of private email review comments
- rebase
- move arch/arm/vdso/vgettimeofday.c to lib/vdso/vgettimeofday.c
- adjust vgettimeofday.c to be a better global candidate, switch to using
ARCH_PROVIDES_TIMER and __arch_counter_get() as more generic.
- do not expose gettimeofday if arch does not support user space timer
v4:
- update commit messages to reflect reasoning here.
- switch to arch_vdso_read_counter as common API.
- split out ARCH_CLOCK_FIXED_MASK.
- Use GENMASK_ULL macro for ARCH_CLOCK_FIXED_MASK.
- dropping forced inline patch and side effects.
- replace all typeof()s with the appropriate types
vdso_wtm_clock_nsec_t, vdso_xtime_clock_sec, vdso_raw_time_sec_t,
long (tv_nsec) and __kernel_time_t (tv_sec).
- clock_getres implementation, use clock instead of clock_id to
match usage elsewhere and remove an inconsistency.
- simplify arm64 implementation of arch_vdso_read_counter,
use read_sysreg.
v5:
- fixes and changes inspired by compiling with clang. Makefile and
portability issues identified. clang build produced a bad arm64
vdso binary until proper compile flags were provided.
- arm compile issue, replace erroneous tk->raw_time.shift with
tk->tkr_raw.shift.
- changes inspired by compat architectural portability, minimize
include files in vgettimeofday.c that notoriously include
unnecessary architectural cruft and defer them to vdso compiler.h.
Problematic direct and indirectly included files are
asm/arch_timer.h, asm/processor.h, asm/sysreg.h, linux/jiffies.h,
linux/hrtimer.h and linux/time.h.
- do not use timespec_add_ns as not accessible via uapi/linux/time.h.
- strip out unused definitions in asm-offsets.c.
From 1583342569880692695@xxx Mon Nov 06 18:32:22 +0000 2017
X-GM-THRID: 1583067986936830066
X-Gmail-Labels: Inbox,Category Forums,HistoricalUnread