From: Christophe Leroy <christophe.leroy@c-s.fr> To: Benjamin Herrenschmidt <benh@kernel.crashing.org>, Paul Mackerras <paulus@samba.org>, Michael Ellerman <mpe@ellerman.id.au>, nathanl@linux.ibm.com Cc: linux-kernel@vger.kernel.org, linuxppc-dev@lists.ozlabs.org, arnd@arndb.de, tglx@linutronix.de, vincenzo.frascino@arm.com, luto@kernel.org, linux-arch@vger.kernel.org Subject: [PATCH v8 7/8] lib/vdso: force inlining of __cvdso_clock_gettime_common() Date: Tue, 28 Apr 2020 13:16:53 +0000 (UTC) [thread overview] Message-ID: <1ab6a62c356c3bec35d1623563ef9c636205bcda.1588079622.git.christophe.leroy@c-s.fr> (raw) In-Reply-To: <cover.1588079622.git.christophe.leroy@c-s.fr> When adding gettime64() to a 32 bit architecture (namely powerpc/32) it has been noticed that GCC doesn't inline anymore __cvdso_clock_gettime_common() because it is called twice (Once by __cvdso_clock_gettime() and once by __cvdso_clock_gettime32). This has the effect of seriously degrading the performance: Before the implementation of gettime64(), gettime() runs in: clock-gettime-monotonic-raw: vdso: 1003 nsec/call clock-gettime-monotonic-coarse: vdso: 592 nsec/call clock-gettime-monotonic: vdso: 942 nsec/call When adding a gettime64() entry point, the standard gettime() performance is degraded by 30% to 50%: clock-gettime-monotonic-raw: vdso: 1300 nsec/call clock-gettime-monotonic-coarse: vdso: 900 nsec/call clock-gettime-monotonic: vdso: 1232 nsec/call Adding __always_inline() to __cvdso_clock_gettime_common() regains the original performance. In terms of code size, the inlining increases the code size by only 176 bytes. This is in the noise for a kernel image. Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr> --- lib/vdso/gettimeofday.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/lib/vdso/gettimeofday.c b/lib/vdso/gettimeofday.c index a2909af4b924..7938d3c4901d 100644 --- a/lib/vdso/gettimeofday.c +++ b/lib/vdso/gettimeofday.c @@ -210,7 +210,7 @@ static __always_inline int do_coarse(const struct vdso_data *vd, clockid_t clk, return 0; } -static __maybe_unused int +static __always_inline int __cvdso_clock_gettime_common(const struct vdso_data *vd, clockid_t clock, struct __kernel_timespec *ts) { -- 2.25.0
WARNING: multiple messages have this Message-ID (diff)
From: Christophe Leroy <christophe.leroy@c-s.fr> To: Benjamin Herrenschmidt <benh@kernel.crashing.org>, Paul Mackerras <paulus@samba.org>, Michael Ellerman <mpe@ellerman.id.au>, nathanl@linux.ibm.com Cc: linux-arch@vger.kernel.org, arnd@arndb.de, linux-kernel@vger.kernel.org, luto@kernel.org, tglx@linutronix.de, vincenzo.frascino@arm.com, linuxppc-dev@lists.ozlabs.org Subject: [PATCH v8 7/8] lib/vdso: force inlining of __cvdso_clock_gettime_common() Date: Tue, 28 Apr 2020 13:16:53 +0000 (UTC) [thread overview] Message-ID: <1ab6a62c356c3bec35d1623563ef9c636205bcda.1588079622.git.christophe.leroy@c-s.fr> (raw) In-Reply-To: <cover.1588079622.git.christophe.leroy@c-s.fr> When adding gettime64() to a 32 bit architecture (namely powerpc/32) it has been noticed that GCC doesn't inline anymore __cvdso_clock_gettime_common() because it is called twice (Once by __cvdso_clock_gettime() and once by __cvdso_clock_gettime32). This has the effect of seriously degrading the performance: Before the implementation of gettime64(), gettime() runs in: clock-gettime-monotonic-raw: vdso: 1003 nsec/call clock-gettime-monotonic-coarse: vdso: 592 nsec/call clock-gettime-monotonic: vdso: 942 nsec/call When adding a gettime64() entry point, the standard gettime() performance is degraded by 30% to 50%: clock-gettime-monotonic-raw: vdso: 1300 nsec/call clock-gettime-monotonic-coarse: vdso: 900 nsec/call clock-gettime-monotonic: vdso: 1232 nsec/call Adding __always_inline() to __cvdso_clock_gettime_common() regains the original performance. In terms of code size, the inlining increases the code size by only 176 bytes. This is in the noise for a kernel image. Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr> --- lib/vdso/gettimeofday.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/lib/vdso/gettimeofday.c b/lib/vdso/gettimeofday.c index a2909af4b924..7938d3c4901d 100644 --- a/lib/vdso/gettimeofday.c +++ b/lib/vdso/gettimeofday.c @@ -210,7 +210,7 @@ static __always_inline int do_coarse(const struct vdso_data *vd, clockid_t clk, return 0; } -static __maybe_unused int +static __always_inline int __cvdso_clock_gettime_common(const struct vdso_data *vd, clockid_t clock, struct __kernel_timespec *ts) { -- 2.25.0
next prev parent reply other threads:[~2020-04-28 13:17 UTC|newest] Thread overview: 71+ messages / expand[flat|nested] mbox.gz Atom feed top 2020-04-28 13:16 [PATCH v8 0/8] powerpc: switch VDSO to C implementation Christophe Leroy 2020-04-28 13:16 ` Christophe Leroy 2020-04-28 13:16 ` [PATCH v8 1/8] powerpc/vdso64: Switch from __get_datapage() to get_datapage inline macro Christophe Leroy 2020-04-28 13:16 ` Christophe Leroy 2020-04-28 13:16 ` [PATCH v8 2/8] powerpc/vdso: Remove __kernel_datapage_offset and simplify __get_datapage() Christophe Leroy 2020-04-28 13:16 ` Christophe Leroy 2020-07-16 2:59 ` Michael Ellerman 2020-07-16 2:59 ` Michael Ellerman 2020-08-04 11:17 ` Christophe Leroy 2020-08-04 11:17 ` Christophe Leroy 2020-08-25 14:15 ` Christophe Leroy 2020-08-26 13:58 ` Michael Ellerman 2020-08-26 13:58 ` Michael Ellerman 2020-08-27 20:34 ` Dmitry Safonov 2020-08-27 20:34 ` Dmitry Safonov 2020-08-28 2:14 ` Michael Ellerman 2020-08-28 2:14 ` Michael Ellerman 2020-09-21 11:26 ` Will Deacon 2020-09-21 11:26 ` Will Deacon 2020-09-27 7:43 ` Christophe Leroy 2020-09-27 7:43 ` Christophe Leroy 2020-09-28 15:08 ` Dmitry Safonov 2020-09-28 15:08 ` Dmitry Safonov 2020-10-23 11:22 ` Christophe Leroy 2020-10-23 11:22 ` Christophe Leroy 2020-10-23 11:25 ` Will Deacon 2020-10-23 11:25 ` Will Deacon 2020-10-23 11:57 ` Christophe Leroy 2020-10-23 11:57 ` Christophe Leroy 2020-10-23 13:29 ` Dmitry Safonov 2020-10-23 13:29 ` Dmitry Safonov 2020-04-28 13:16 ` [PATCH v8 3/8] powerpc/vdso: Remove unused \tmp param in __get_datapage() Christophe Leroy 2020-04-28 13:16 ` Christophe Leroy 2020-04-28 13:16 ` [PATCH v8 4/8] powerpc/processor: Move cpu_relax() into asm/vdso/processor.h Christophe Leroy 2020-04-28 13:16 ` Christophe Leroy 2020-04-28 13:16 ` [PATCH v8 5/8] powerpc/vdso: Prepare for switching VDSO to generic C implementation Christophe Leroy 2020-04-28 13:16 ` Christophe Leroy 2020-07-15 1:04 ` Michael Ellerman 2020-07-15 1:04 ` Michael Ellerman 2020-07-15 18:47 ` Christophe Leroy 2020-07-15 18:47 ` Christophe Leroy 2020-07-16 23:18 ` Tulio Magno Quites Machado Filho 2020-07-16 23:18 ` Tulio Magno Quites Machado Filho 2020-08-04 11:14 ` Christophe Leroy 2020-08-04 11:14 ` Christophe Leroy 2020-08-05 6:24 ` Michael Ellerman 2020-08-05 6:24 ` Michael Ellerman 2020-08-05 13:35 ` Segher Boessenkool 2020-08-05 13:35 ` Segher Boessenkool 2020-08-06 2:03 ` Michael Ellerman 2020-08-06 2:03 ` Michael Ellerman 2020-08-06 18:33 ` Segher Boessenkool 2020-08-06 18:33 ` Segher Boessenkool 2020-08-07 2:44 ` Michael Ellerman 2020-08-07 2:44 ` Michael Ellerman 2020-04-28 13:16 ` [PATCH v8 6/8] powerpc/vdso: Switch " Christophe Leroy 2020-04-28 13:16 ` Christophe Leroy 2020-04-28 13:16 ` Christophe Leroy [this message] 2020-04-28 13:16 ` [PATCH v8 7/8] lib/vdso: force inlining of __cvdso_clock_gettime_common() Christophe Leroy 2020-04-28 13:16 ` [PATCH v8 8/8] powerpc/vdso: Provide __kernel_clock_gettime64() on vdso32 Christophe Leroy 2020-04-28 13:16 ` Christophe Leroy 2020-04-28 15:03 ` Christophe Leroy 2020-04-28 16:05 ` Arnd Bergmann 2020-04-28 16:05 ` Arnd Bergmann 2020-05-09 15:54 ` Christophe Leroy 2020-05-09 15:54 ` Christophe Leroy 2020-05-09 18:48 ` Christophe Leroy 2020-05-29 18:56 ` [PATCH v8 0/8] powerpc: switch VDSO to C implementation Christophe Leroy 2020-06-03 10:04 ` Michael Ellerman 2020-07-16 12:55 ` Michael Ellerman 2020-07-16 12:55 ` Michael Ellerman
Reply instructions: You may reply publicly to this message via plain-text email using any one of the following methods: * Save the following mbox file, import it into your mail client, and reply-to-all from there: mbox Avoid top-posting and favor interleaved quoting: https://en.wikipedia.org/wiki/Posting_style#Interleaved_style * Reply using the --to, --cc, and --in-reply-to switches of git-send-email(1): git send-email \ --in-reply-to=1ab6a62c356c3bec35d1623563ef9c636205bcda.1588079622.git.christophe.leroy@c-s.fr \ --to=christophe.leroy@c-s.fr \ --cc=arnd@arndb.de \ --cc=benh@kernel.crashing.org \ --cc=linux-arch@vger.kernel.org \ --cc=linux-kernel@vger.kernel.org \ --cc=linuxppc-dev@lists.ozlabs.org \ --cc=luto@kernel.org \ --cc=mpe@ellerman.id.au \ --cc=nathanl@linux.ibm.com \ --cc=paulus@samba.org \ --cc=tglx@linutronix.de \ --cc=vincenzo.frascino@arm.com \ /path/to/YOUR_REPLY https://kernel.org/pub/software/scm/git/docs/git-send-email.html * If your mail client supports setting the In-Reply-To header via mailto: links, try the mailto: linkBe sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.