From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from simark.ca by simark.ca with LMTP id 3qSVCZTk0mPoTCIAWB0awg (envelope-from ) for ; Thu, 26 Jan 2023 15:37:40 -0500 Received: by simark.ca (Postfix, from userid 112) id 19F9B1E128; Thu, 26 Jan 2023 15:37:40 -0500 (EST) Authentication-Results: simark.ca; dkim=pass (2048-bit key; unprotected) header.d=lists.lttng.org header.i=@lists.lttng.org header.a=rsa-sha256 header.s=default header.b=D0+MrIOU; dkim-atps=neutral X-Spam-Checker-Version: SpamAssassin 3.4.6 (2021-04-09) on simark.ca X-Spam-Level: X-Spam-Status: No, score=-2.9 required=5.0 tests=BAYES_00,DKIM_SIGNED, DKIM_VALID,DKIM_VALID_AU,HTML_MESSAGE,MAILING_LIST_MULTI, MIME_HTML_MOSTLY,URIBL_BLOCKED autolearn=ham autolearn_force=no version=3.4.6 Received: from lists.lttng.org (lists.lttng.org [167.114.26.123]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange X25519 server-signature RSA-PSS (2048 bits)) (No client certificate requested) by simark.ca (Postfix) with ESMTPS id 091161E110 for ; Thu, 26 Jan 2023 15:37:39 -0500 (EST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=lists.lttng.org; s=default; t=1674765458; bh=am41u330ZAFhcS5Z+OcAEIgHuu93kgyp1/QZR4H6jWI=; h=References:In-Reply-To:Date:To:Cc:Subject:List-Id: List-Unsubscribe:List-Archive:List-Post:List-Help:List-Subscribe: From:Reply-To:From; b=D0+MrIOUwxtq0ktlwaykhgjWv+ejYzf2TE5pIp2ac7Fb0E5gPOiq4bCvQz/OOQYhD ySFcBCcfBOLsCkXP4pWXNOkDlgOJIfs1DFm9aD2oB3bBcGqbAk2HtVCtCNWUy/Y3vc eEp+Z8/lAB/N017cVhyVQf87egLnVFjZ+Qo3Cjiv+ttnOty3L3a7i5SCB8ltQfMaAX hzCE+RE93+b7WPjw8Zpx+yH38m/3cfGtaMo4KWSsoBQ2dRHnfecZ9oc6/wglnqxwd7 U01qzpIzw829qhjThpib7Iqolzu7Fd8yFjLHHQ+NlgNr11q+sg/kNSyEkm8qGb1Xb5 LLPfbZq+sa4UQ== Received: from lists-lttng01.efficios.com (localhost [IPv6:::1]) by lists.lttng.org (Postfix) with ESMTP id 4P2sw224F8z1Vny; Thu, 26 Jan 2023 15:37:38 -0500 (EST) Received: from mail-pg1-f182.google.com (mail-pg1-f182.google.com [209.85.215.182]) by lists.lttng.org (Postfix) with ESMTPS id 4P2rTl1vGvz1W2q for ; Thu, 26 Jan 2023 14:33:15 -0500 (EST) Received: by mail-pg1-f182.google.com with SMTP id 141so1796040pgc.0 for ; Thu, 26 Jan 2023 11:33:15 -0800 (PST) X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20210112; h=cc:to:subject:message-id:date:from:in-reply-to:references :mime-version:x-gm-message-state:from:to:cc:subject:date:message-id :reply-to; bh=KyzL9pjEmU8wm4DHHs28TRUPUHXT8A7XQbiaHa50W8E=; b=AQgYGhm7Ye1XVGKK0mEZMMT18GwutrD0fkW8wWyj0Nty6TFwkFnLz6L6D/tnjTm/+U zMcXazj2mI60BgsVQDo+MPqMIHMAoNfA3zLMstQm7Sg3jyi/EFaO3UB7N53cn+8PFdCp OICbY0pKpEHJNFzMAHnh82YZ4vmsoo9IiVZif3V118xikJuP0ajgTgtH53Ao901HT1qe wNVZ4+W0Xx+0ThBUq7QOwjSWf03go95PY0GOCIfqLgaBZvQrbG9c93Oo+Fn+BByvon0j xr2JnsnTa3yX1EZN/WUAK8EuE0VJJyuO5VaZtDVf9U0YGILDjbWpXnt9DxV3bUz5394f e7Tg== X-Gm-Message-State: AFqh2koixscfxSV60EQkzeoZFH958WKhwzAsDwf9fd4uzL7O5RLCZn7p SoHjcvyLXkQ2JOZT9jsk7ZzYA6DU7aK9WsP1w4I= X-Google-Smtp-Source: AMrXdXsD2n1qo1LH8LCk4sl8ObdgcWNzZy5TAuZDtkAm63ScNosuiJAkp3hdKgenrf3XAeT11lS/dggJVAj+9qUVCpA= X-Received: by 2002:a63:f017:0:b0:477:bfab:5a81 with SMTP id k23-20020a63f017000000b00477bfab5a81mr3690578pgh.64.1674761533490; Thu, 26 Jan 2023 11:32:13 -0800 (PST) MIME-Version: 1.0 References: <954909194.21985.1511551181212.JavaMail.zimbra@efficios.com> <1784f4f4-1c5a-5c4b-2041-2327128712d1@efficios.com> In-Reply-To: <1784f4f4-1c5a-5c4b-2041-2327128712d1@efficios.com> Date: Thu, 26 Jan 2023 20:32:01 +0100 Message-ID: To: Mathieu Desnoyers Cc: lttng-dev X-Mailman-Approved-At: Thu, 26 Jan 2023 15:37:37 -0500 Subject: Re: [lttng-dev] lttng-consumerd crash on aarch64 due to x86 arch specific optimization X-BeenThere: lttng-dev@lists.lttng.org X-Mailman-Version: 2.1.39 Precedence: list List-Id: LTTng development list List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , From: Anders Wallin via lttng-dev Reply-To: Anders Wallin Content-Type: multipart/mixed; boundary="===============6930845206339412935==" Errors-To: lttng-dev-bounces@lists.lttng.org Sender: "lttng-dev" --===============6930845206339412935== Content-Type: multipart/alternative; boundary="000000000000498caf05f32fcef9" --000000000000498caf05f32fcef9 Content-Type: text/plain; charset="UTF-8" Hi Matthieu, I've retired and no longer have access to any arch64 target to test it on. Regards Anders Den ons 25 jan. 2023 13:25Mathieu Desnoyers skrev: > Hi Anders, > > Sorry for the long delay on this one, can you have a look at the following > fix ? > > https://review.lttng.org/c/lttng-ust/+/9319 Fix: aarch64: do not perform > unaligned stores > > If it passes your testing, I'll merge this into lttng-ust. > > Thanks, > > Mathieu > > On 2017-12-28 09:13, Anders Wallin wrote: > > Hi Mathieu, > > > > I finally got some time to dig into this issue. The crash only happens > > when metadata is written AND the size of the metadata will end up in a > > write that is 8,4,2 or 1 bytes long AND > > that the source or destination is not aligned correctly according to HW > > limitation. I have not found any simple way to keep the performance > > enhancement code that is run most of the time. > > Maybe the metadata writes should have it's own write function instead. > > > > Here is an example of a crash (code is from lttng-ust 2.9.1 and > > lttng-tools 2.9.6) where the size is 8 bytes and the src address is > > unaligned at 0xf3b7eeb2; > > > > #0 lttng_inline_memcpy (len=8, src=0xf3b7eeb2, dest=) at > > /usr/src/debug/lttng-ust/2.9.1/git/libringbuffer/backend_internal.h:610 > > No locals. > > #1 lib_ring_buffer_write (len=8, src=0xf3b7eeb2, ctx=0xf57c47d0, > > config=0xf737c560 ) at > > /usr/src/debug/lttng-ust/2.9.1/git/libringbuffer/backend.h:100 > > __len = 8 > > handle = 0xf3b2e0c0 > > backend_pages = > > chanb = 0xf3b2e2e0 > > offset = > > > > #2 lttng_event_write (ctx=0xf57c47d0, src=0xf3b7eeb2, len=8) at > > > /usr/src/debug/lttng-ust/2.9.1/git/liblttng-ust/lttng-ring-buffer-metadata-client.h:267 > > No locals. > > > > #3 0xf7337ef8 in ustctl_write_one_packet_to_channel (channel= > out>, metadata_str=0xf3b7eeb2 "", len=) at > > /usr/src/debug/lttng-ust/2.9.1/git/liblttng-ust-ctl/ustctl.c:1183 > > ctx = {chan = 0xf3b2e290, priv = 0x0, handle = 0xf3b2e0c0, > > data_size = 8, largest_align = 1, cpu = -1, buf = 0xf6909000, slot_size > > = 8, buf_offset = 163877, pre_offset = 163877, tsc = 0, rflags = 0, > > ctx_len = 80, ip = 0x0, priv2 = 0x0, padding2 = '\000' > times>, backend_pages = 0xf690c000} > > chan = 0xf3b2e4d8 > > str = 0xf3b7eeb2 "" > > reserve_len = 8 > > ret = > > __func__ = '\000' > > __PRETTY_FUNCTION__ = '\000' > > ---Type to continue, or q to quit--- > > > > #4 0x000344cc in commit_one_metadata_packet > > (stream=stream@entry=0xf3b2e560) at ust-consumer.c:2206 > > write_len = > > ret = > > __PRETTY_FUNCTION__ = "commit_one_metadata_packet" > > > > #5 0x00036538 in lttng_ustconsumer_read_subbuffer > > (stream=stream@entry=0xf3b2e560, ctx=ctx@entry=0x25e6e8) at > > ust-consumer.c:2452 > > len = 4096 > > subbuf_size = 4093 > > padding = > > err = -11 > > write_index = 1 > > ret = > > ustream = > > index = {offset = 0, packet_size = 575697416355872, > > content_size = 17564043391468256584, timestamp_begin = > > 17564043425827782792, timestamp_end = 34359738496, > > Regards > > Anders > > > > fre 24 nov. 2017 kl 20:18 skrev Mathieu Desnoyers > > >>: > > > > ----- On Nov 24, 2017, at 3:23 AM, Anders Wallin > > wrote: > > > > Hi, > > architectures that has memory alignment restrictions may/will > > fail with the > > optimization done in 51b8f2fa2b972e62117caa946dd3e3565b6ca4a3. > > Please revert the patch or make it X86 specific. > > > > > > Hi Anders, > > > > This was added in the development cycle of lttng-ust 2.9. We could > > perhaps > > add a test on the pointer alignment for architectures that care > > about it, and > > fallback to memcpy in those cases. > > > > The revert approach would have been justified if this commit had > > been backported > > as a "fix" to a stable branch, which is not the case here. We should > > work on > > finding an acceptable solution that takes care of dealing with > > unaligned pointers > > on architectures that care about the difference. > > > > Thanks, > > > > Mathieu > > > > > > > > Regards > > > > Anders Wallin > > > ------------------------------------------------------------------------------------------------------------ > > commit 51b8f2fa2b972e62117caa946dd3e3565b6ca4a3 > > Author: Mathieu Desnoyers > > > > Date: Sun Sep 25 12:31:11 2016 -0400 > > > > Performance: implement lttng_inline_memcpy > > Because all length parameters received for serializing data > > coming from > > applications go through a callback, they are never > > constant, and it > > hurts performance to perform a call to memcpy each time. > > Signed-off-by: Mathieu Desnoyers > > > > > > > > diff --git a/libringbuffer/backend_internal.h > > b/libringbuffer/backend_internal.h > > index 90088b89..e597cf4d 100644 > > --- a/libringbuffer/backend_internal.h > > +++ b/libringbuffer/backend_internal.h > > @@ -592,6 +592,28 @@ int update_read_sb_index(const struct > > lttng_ust_lib_ring_buffer_config *config, > > #define inline_memcpy(dest, src, n) memcpy(dest, src, n) > > #endif > > +static inline __attribute__((always_inline)) > > +void lttng_inline_memcpy(void *dest, const void *src, > > + unsigned long len) > > +{ > > + switch (len) { > > + case 1: > > + *(uint8_t *) dest = *(const uint8_t *) src; > > + break; > > + case 2: > > + *(uint16_t *) dest = *(const uint16_t *) src; > > + break; > > + case 4: > > + *(uint32_t *) dest = *(const uint32_t *) src; > > + break; > > + case 8: > > + *(uint64_t *) dest = *(const uint64_t *) src; > > + break; > > + default: > > + inline_memcpy(dest, src, len); > > + } > > +} > > + > > /* > > * Use the architecture-specific memcpy implementation for > > constant-sized > > * inputs, but rely on an inline memcpy for length statically > > unknown. > > @@ -603,7 +625,7 @@ do { > > \ > > if (__builtin_constant_p(len)) > \ > > memcpy(dest, src, __len); > \ > > else > \ > > - inline_memcpy(dest, src, __len); \ > > + lttng_inline_memcpy(dest, src, __len); \ > > } while (0) > > /* > > > > > ---------------------------------------------------------------------------------------------------- > > Here is one example of a crash, where 0xf3b7eeb2 is not 8 byte > > aligned; > > > > (gdb) bt full > > #0 lttng_inline_memcpy (len=8, src=0xf3b7eeb2, dest= > out>) at > > > /usr/src/debug/lttng-ust/2.9.1/git/libringbuffer/backend_internal.h:610 > > No locals. > > #1 lib_ring_buffer_write (len=8, src=0xf3b7eeb2, > > ctx=0xf57c47d0, config=0xf737c560 ) at > > /usr/src/debug/lttng-ust/2.9.1/git/libringbuffer/backend.h:100 > > __len = 8 > > handle = 0xf3b2e0c0 > > backend_pages = > > chanb = 0xf3b2e2e0 > > offset = > > #2 lttng_event_write (ctx=0xf57c47d0, src=0xf3b7eeb2, len=8) at > > > /usr/src/debug/lttng-ust/2.9.1/git/liblttng-ust/lttng-ring-buffer-metadata-client.h:267 > > No locals. > > #3 0xf7337ef8 in ustctl_write_one_packet_to_channel > > (channel=, metadata_str=0xf3b7eeb2 "", > > len=) at > > /usr/src/debug/lttng-ust/2.9.1/git/liblttng-ust-ctl/ustctl.c:1183 > > ctx = {chan = 0xf3b2e290, priv = 0x0, handle = > > 0xf3b2e0c0, data_size = 8, largest_align = 1, cpu = -1, buf = > > 0xf6909000, slot_size = 8, buf_offset = 163877, pre_offset = > > 163877, tsc = 0, rflags = 0, ctx_len = 80, ip = 0x0, priv2 = > > 0x0, padding2 = '\000' , backend_pages = > > 0xf690c000} > > chan = 0xf3b2e4d8 > > str = 0xf3b7eeb2 "" > > reserve_len = 8 > > ret = > > __func__ = '\000' > > __PRETTY_FUNCTION__ = '\000' > > ---Type to continue, or q to quit--- > > #4 0x000344cc in commit_one_metadata_packet > > (stream=stream@entry=0xf3b2e560) at ust-consumer.c:2206 > > write_len = > > ret = > > __PRETTY_FUNCTION__ = "commit_one_metadata_packet" > > #5 0x00036538 in lttng_ustconsumer_read_subbuffer > > (stream=stream@entry=0xf3b2e560, ctx=ctx@entry=0x25e6e8) at > > ust-consumer.c:2452 > > len = 4096 > > subbuf_size = 4093 > > padding = > > err = -11 > > write_index = 1 > > ret = > > ustream = > > index = {offset = 0, packet_size = 575697416355872, > > content_size = 17564043391468256584, timestamp_begin = > > 17564043425827782792, timestamp_end = 34359738496, > > events_discarded = 313327092840, stream_id = 4089446416, > > stream_instance_id = 17689097291346376608, packet_seq_num = > > 8589934593} > > __PRETTY_FUNCTION__ = "lttng_ustconsumer_read_subbuffer" > > __func__ = "lttng_ustconsumer_read_subbuffer" > > ..... > > > > _______________________________________________ > > lttng-dev mailing list > > lttng-dev@lists.lttng.org > > https://lists.lttng.org/cgi-bin/mailman/listinfo/lttng-dev > > > > > > > > -- > > Mathieu Desnoyers > > EfficiOS Inc. > > http://www.efficios.com > > > > -- > Mathieu Desnoyers > EfficiOS Inc. > https://www.efficios.com > > --000000000000498caf05f32fcef9 Content-Type: text/html; charset="UTF-8" Content-Transfer-Encoding: quoted-printable
Hi Matthieu,

I've retired and no longer have access to any arch64=C2=A0 target= to test it on.=C2=A0

Re= gards
Anders


Den ons 25 jan. 2023 13:25Mathieu De= snoyers <mathieu.desno= yers@efficios.com> skrev:
Hi= Anders,

Sorry for the long delay on this one, can you have a look at the following = fix ?

https://review.lttng.org/c/lttng-ust/+/9319 Fix: aarch64: do not perform unaligned stores

If it passes your testing, I'll merge this into lttng-ust.

Thanks,

Mathieu

On 2017-12-28 09:13, Anders Wallin wrote:
> Hi Mathieu,
>
> I finally got some time to dig into this issue. The crash only happens=
> when metadata is written AND the size of the metadata will end up in a=
> write that is 8,4,2 or 1 bytes long AND
> that the source or destination is not aligned correctly according to H= W
> limitation. I have not found any simple way to keep the performance > enhancement code that is run most of the time.
> Maybe the metadata writes should have it's own write function inst= ead.
>
> Here is an example of a crash (code is from lttng-ust 2.9.1 and
> lttng-tools 2.9.6) where the size is 8 bytes and the src address is > unaligned at 0xf3b7eeb2;
>
> #0=C2=A0 lttng_inline_memcpy (len=3D8, src=3D0xf3b7eeb2, dest=3D<op= timized out>) at
> /usr/src/debug/lttng-ust/2.9.1/git/libringbuffer/backend_internal.h:61= 0
> No locals.
> #1=C2=A0 lib_ring_buffer_write (len=3D8, src=3D0xf3b7eeb2, ctx=3D0xf57= c47d0,
> config=3D0xf737c560 <client_config>) at
> /usr/src/debug/lttng-ust/2.9.1/git/libringbuffer/backend.h:100
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 __len =3D 8
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 handle =3D 0xf3b2e0c0
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 backend_pages =3D <optimized out&= gt;
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 chanb =3D 0xf3b2e2e0
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 offset =3D <optimized out>
>
> #2=C2=A0 lttng_event_write (ctx=3D0xf57c47d0, src=3D0xf3b7eeb2, len=3D= 8) at
> /usr/src/debug/lttng-ust/2.9.1/git/liblttng-ust/lttng-ring-buffer-meta= data-client.h:267
> No locals.
>
> #3=C2=A0 0xf7337ef8 in ustctl_write_one_packet_to_channel (channel=3D&= lt;optimized
> out>, metadata_str=3D0xf3b7eeb2 "", len=3D<optimized o= ut>) at
> /usr/src/debug/lttng-ust/2.9.1/git/liblttng-ust-ctl/ustctl.c:1183
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 ctx =3D {chan =3D 0xf3b2e290, priv = =3D 0x0, handle =3D 0xf3b2e0c0,
> data_size =3D 8, largest_align =3D 1, cpu =3D -1, buf =3D 0xf6909000, = slot_size
> =3D 8, buf_offset =3D 163877, pre_offset =3D 163877, tsc =3D 0, rflags= =3D 0,
> ctx_len =3D 80, ip =3D 0x0, priv2 =3D 0x0, padding2 =3D '\000'= <repeats 11
> times>, backend_pages =3D 0xf690c000}
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 chan =3D 0xf3b2e4d8
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 str =3D 0xf3b7eeb2 ""
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 reserve_len =3D 8
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 ret =3D <optimized out>
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 __func__ =3D '\000' <repe= ats 34 times>
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 __PRETTY_FUNCTION__ =3D '\000= 9; <repeats 34 times>
> ---Type <return> to continue, or q <return> to quit---
>
> #4=C2=A0 0x000344cc in commit_one_metadata_packet
> (stream=3Dstream@entry=3D0xf3b2e560) at ust-consumer.c:2206
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 write_len =3D <optimized out><= br> >=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 ret =3D <optimized out>
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 __PRETTY_FUNCTION__ =3D "commit= _one_metadata_packet"
>
> #5=C2=A0 0x00036538 in lttng_ustconsumer_read_subbuffer
> (stream=3Dstream@entry=3D0xf3b2e560, ctx=3Dctx@entry=3D0x25e6e8) at > ust-consumer.c:2452
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 len =3D 4096
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 subbuf_size =3D 4093
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 padding =3D <optimized out> >=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 err =3D -11
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 write_index =3D 1
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 ret =3D <optimized out>
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 ustream =3D <optimized out> >=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 index =3D {offset =3D 0, packet_size= =3D 575697416355872,
> content_size =3D 17564043391468256584, timestamp_begin =3D
> 17564043425827782792, timestamp_end =3D 34359738496,
> Regards
> Anders
>
> fre 24 nov. 2017 kl 20:18 skrev Mathieu Desnoyers
> <mathieu.desnoyers@efficios.com <mailto:mathieu.desnoyers@efficios.com>>:
>
>=C2=A0 =C2=A0 =C2=A0----- On Nov 24, 2017, at 3:23 AM, Anders Wallin &l= t;wallinux@gmail.com
>=C2=A0 =C2=A0 =C2=A0<mailto:wallinux@gmail.com>> wrote: >
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0Hi,
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0architectures that has memory alignme= nt restrictions may/will
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0fail with the
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0optimization done in=C2=A051b8f2fa2b9= 72e62117caa946dd3e3565b6ca4a3.
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0Please revert the patch or make it X8= 6 specific.
>
>
>=C2=A0 =C2=A0 =C2=A0Hi Anders,
>
>=C2=A0 =C2=A0 =C2=A0This was added in the development cycle of lttng-us= t 2.9. We could
>=C2=A0 =C2=A0 =C2=A0perhaps
>=C2=A0 =C2=A0 =C2=A0add a test on the pointer alignment for architectur= es that care
>=C2=A0 =C2=A0 =C2=A0about it, and
>=C2=A0 =C2=A0 =C2=A0fallback to memcpy in those cases.
>
>=C2=A0 =C2=A0 =C2=A0The revert approach would have been justified if th= is commit had
>=C2=A0 =C2=A0 =C2=A0been backported
>=C2=A0 =C2=A0 =C2=A0as a "fix" to a stable branch, which is n= ot the case here. We should
>=C2=A0 =C2=A0 =C2=A0work on
>=C2=A0 =C2=A0 =C2=A0finding an acceptable solution that takes care of d= ealing with
>=C2=A0 =C2=A0 =C2=A0unaligned pointers
>=C2=A0 =C2=A0 =C2=A0on architectures that care about the difference. >
>=C2=A0 =C2=A0 =C2=A0Thanks,
>
>=C2=A0 =C2=A0 =C2=A0Mathieu
>
>
>
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0Regards
>
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0Anders Wallin
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0-------------------------------------= -----------------------------------------------------------------------
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0commit 51b8f2fa2b972e62117caa946dd3e3= 565b6ca4a3
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0Author: Mathieu Desnoyers <mathieu.desnoyers@efficios.com
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0<mailto:mathieu.desnoy= ers@efficios.com>>
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0Date:=C2=A0 =C2=A0Sun Sep 25 12:31:11= 2016 -0400
>
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 Performance: implement= lttng_inline_memcpy
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 Because all length par= ameters received for serializing data
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0coming from
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 applications go throug= h a callback, they are never
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0constant, and it
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 hurts performance to p= erform a call to memcpy each time.
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 Signed-off-by: Mathieu= Desnoyers
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0<mathieu.desnoyers@eff= icios.com
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0<mailto:mathieu.desnoy= ers@efficios.com>>
>
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0diff --git a/libringbuffer/backend_in= ternal.h
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0b/libringbuffer/backend_internal.h >=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0index 90088b89..e597cf4d 100644
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0--- a/libringbuffer/backend_internal.= h
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0+++ b/libringbuffer/backend_internal.= h
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0@@ -592,6 +592,28 @@ int update_read_= sb_index(const struct
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0lttng_ust_lib_ring_buffer_config *con= fig,
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0#define inline_memcpy(dest, sr= c, n)=C2=A0 =C2=A0 memcpy(dest, src, n)
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0#endif
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0+static inline __attribute__((always_= inline))
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0+void lttng_inline_memcpy(void *dest,= const void *src,
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0+=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 = =C2=A0 =C2=A0 =C2=A0unsigned long len)
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0+{
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0+=C2=A0 =C2=A0 =C2=A0 =C2=A0switch (l= en) {
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0+=C2=A0 =C2=A0 =C2=A0 =C2=A0case 1: >=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0+=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 = =C2=A0 =C2=A0 =C2=A0*(uint8_t *) dest =3D *(const uint8_t *) src;
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0+=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 = =C2=A0 =C2=A0 =C2=A0break;
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0+=C2=A0 =C2=A0 =C2=A0 =C2=A0case 2: >=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0+=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 = =C2=A0 =C2=A0 =C2=A0*(uint16_t *) dest =3D *(const uint16_t *) src;
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0+=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 = =C2=A0 =C2=A0 =C2=A0break;
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0+=C2=A0 =C2=A0 =C2=A0 =C2=A0case 4: >=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0+=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 = =C2=A0 =C2=A0 =C2=A0*(uint32_t *) dest =3D *(const uint32_t *) src;
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0+=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 = =C2=A0 =C2=A0 =C2=A0break;
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0+=C2=A0 =C2=A0 =C2=A0 =C2=A0case 8: >=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0+=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 = =C2=A0 =C2=A0 =C2=A0*(uint64_t *) dest =3D *(const uint64_t *) src;
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0+=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 = =C2=A0 =C2=A0 =C2=A0break;
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0+=C2=A0 =C2=A0 =C2=A0 =C2=A0default:<= br> >=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0+=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 = =C2=A0 =C2=A0 =C2=A0inline_memcpy(dest, src, len);
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0+=C2=A0 =C2=A0 =C2=A0 =C2=A0}
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0+}
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0+
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0/*
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 * Use the architecture-specif= ic memcpy implementation for
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0constant-sized
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 * inputs, but rely on an inli= ne memcpy for length statically
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0unknown.
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0@@ -603,7 +625,7 @@ do {=C2=A0 =C2=A0= =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2= =A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 = =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 \
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 if (__bu= iltin_constant_p(len))=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2= =A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 \
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 = =C2=A0 =C2=A0 =C2=A0 memcpy(dest, src, __len);=C2=A0 =C2=A0 =C2=A0 =C2=A0 = =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0\
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 else=C2= =A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 = =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2= =A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 \
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0-=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 = =C2=A0 =C2=A0 =C2=A0inline_memcpy(dest, src, __len);=C2=A0 =C2=A0 =C2=A0 = =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 \
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0+=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 = =C2=A0 =C2=A0 =C2=A0lttng_inline_memcpy(dest, src, __len);=C2=A0 =C2=A0 =C2= =A0 =C2=A0 =C2=A0 \
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0} while (0)
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0/*
>
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0-------------------------------------= ---------------------------------------------------------------
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0Here is one example of a crash, where= 0xf3b7eeb2 is not 8 byte
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0aligned;
>
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 (gdb) bt full
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0#0=C2=A0 lttng_inline_memcpy (len=3D8= , src=3D0xf3b7eeb2, dest=3D<optimized
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0out>) at
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0/usr/src/debug/lttng-ust/2.9.1/git/li= bringbuffer/backend_internal.h:610
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0No locals.
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0#1=C2=A0 lib_ring_buffer_write (len= =3D8, src=3D0xf3b7eeb2,
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0ctx=3D0xf57c47d0, config=3D0xf737c560= <client_config>) at
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0/usr/src/debug/lttng-ust/2.9.1/git/li= bringbuffer/backend.h:100
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 __len = =3D 8
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 handle = =3D 0xf3b2e0c0
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 backend_= pages =3D <optimized out>
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 chanb = =3D 0xf3b2e2e0
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 offset = =3D <optimized out>
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0#2=C2=A0 lttng_event_write (ctx=3D0xf= 57c47d0, src=3D0xf3b7eeb2, len=3D8) at
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0/usr/src/debug/lttng-ust/2.9.1/git/li= blttng-ust/lttng-ring-buffer-metadata-client.h:267
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0No locals.
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0#3=C2=A0 0xf7337ef8 in ustctl_write_o= ne_packet_to_channel
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0(channel=3D<optimized out>, met= adata_str=3D0xf3b7eeb2 "",
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0len=3D<optimized out>) at
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0/usr/src/debug/lttng-ust/2.9.1/git/li= blttng-ust-ctl/ustctl.c:1183
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 ctx =3D = {chan =3D 0xf3b2e290, priv =3D 0x0, handle =3D
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A00xf3b2e0c0, data_size =3D 8, largest_= align =3D 1, cpu =3D -1, buf =3D
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A00xf6909000, slot_size =3D 8, buf_offs= et =3D 163877, pre_offset =3D
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0163877, tsc =3D 0, rflags =3D 0, ctx_= len =3D 80, ip =3D 0x0, priv2 =3D
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A00x0, padding2 =3D '\000' <= repeats 11 times>, backend_pages =3D
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A00xf690c000}
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 chan =3D= 0xf3b2e4d8
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 str =3D = 0xf3b7eeb2 ""
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 reserve_= len =3D 8
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 ret =3D = <optimized out>
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 __func__= =3D '\000' <repeats 34 times>
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 __PRETTY= _FUNCTION__ =3D '\000' <repeats 34 times>
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0---Type <return> to continue, o= r q <return> to quit---
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0#4=C2=A0 0x000344cc in commit_one_met= adata_packet
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0(stream=3Dstream@entry=3D0xf3b2e560) = at ust-consumer.c:2206
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 write_le= n =3D <optimized out>
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 ret =3D = <optimized out>
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 __PRETTY= _FUNCTION__ =3D "commit_one_metadata_packet"
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0#5=C2=A0 0x00036538 in lttng_ustconsu= mer_read_subbuffer
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0(stream=3Dstream@entry=3D0xf3b2e560, = ctx=3Dctx@entry=3D0x25e6e8) at
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0ust-consumer.c:2452
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 len =3D = 4096
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 subbuf_s= ize =3D 4093
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 padding = =3D <optimized out>
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 err =3D = -11
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 write_in= dex =3D 1
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 ret =3D = <optimized out>
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 ustream = =3D <optimized out>
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 index = =3D {offset =3D 0, packet_size =3D 575697416355872,
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0content_size =3D 17564043391468256584= , timestamp_begin =3D
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A017564043425827782792, timestamp_end = =3D 34359738496,
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0events_discarded =3D 313327092840, st= ream_id =3D 4089446416,
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0stream_instance_id =3D 17689097291346= 376608, packet_seq_num =3D
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A08589934593}
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 __PRETTY= _FUNCTION__ =3D "lttng_ustconsumer_read_subbuffer"
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 __func__= =3D "lttng_ustconsumer_read_subbuffer"
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0.....
>
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0_____________________________________= __________
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0lttng-dev mailing list
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0lttng-dev@lists.lttng.org = <mailto:lttng-dev@lists.lttng.org>
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0https://lists.lttng.org/cgi-bin/mailman/listinfo/lttng-dev
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0<https://lists.lttng.org/cgi-bin/mailman/listinfo/lttng-dev&= gt;
>
>
>=C2=A0 =C2=A0 =C2=A0--
>=C2=A0 =C2=A0 =C2=A0Mathieu Desnoyers
>=C2=A0 =C2=A0 =C2=A0EfficiOS Inc.
>=C2=A0 =C2=A0 =C2=A0http://www.efficios.com <h= ttp://www.efficios.com>
>

--
Mathieu Desnoyers
EfficiOS Inc.
https://www.efficios.com

--000000000000498caf05f32fcef9-- --===============6930845206339412935== Content-Type: text/plain; charset="us-ascii" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Content-Disposition: inline _______________________________________________ lttng-dev mailing list lttng-dev@lists.lttng.org https://lists.lttng.org/cgi-bin/mailman/listinfo/lttng-dev --===============6930845206339412935==--