From: Jerin Jacob <jerin.jacob@caviumnetworks.com>
To: Gavin Hu <Gavin.Hu@arm.com>
Cc: "dev@dpdk.org" <dev@dpdk.org>,
"Jacob, Jerin" <Jerin.JacobKollanukkaran@cavium.com>
Subject: Re: [dpdk-dev] dev Digest, Vol 195, Issue 45
Date: Wed, 16 May 2018 10:35:06 +0530 [thread overview]
Message-ID: <20180516050505.GA16248@jerin> (raw)
In-Reply-To: <DB7PR08MB31637C042CB07826BE425E5F8F920@DB7PR08MB3163.eurprd08.prod.outlook.com>
-----Original Message-----
> Date: Wed, 16 May 2018 02:08:48 +0000
> From: Gavin Hu <Gavin.Hu@arm.com>
> To: "dev@dpdk.org" <dev@dpdk.org>, "Jacob, Jerin"
> <Jerin.JacobKollanukkaran@cavium.com>
> Subject: Re: [dpdk-dev] dev Digest, Vol 195, Issue 45
>
> >
> > For ARM64 platform, libdpdk.a includes the
> > librte_pmd_octeontx_ssovf.a, which requires the libatomic.a
> > support.The atomic lib is built-in in the gcc toolchain, but for clang it has to be explicitly linked.
> > For more details, please refer to
> > https://clang.llvm.org/docs/Toolchain.html
> >
> > ~/dpdk/build/lib/librte_pmd_octeontx_ssovf.a(timvf_worker.o): In function `timvf_timer_cancel_burst':
> > timvf_worker.c:(.text+0x80): undefined reference to `__atomic_fetch_add_8'
> > /home/gavin/arm_repo/dpdk/build/lib/librte_pmd_octeontx_ssovf.a(timvf_worker.o): In function `timvf_timer_arm_burst_sp':
> > timvf_worker.c:(.text+0x200): undefined reference to `__atomic_fetch_add_8'
> > timvf_worker.c:(.text+0x244): undefined reference to `__atomic_store_2'
> > timvf_worker.c:(.text+0x278): undefined reference to `__atomic_fetch_add_4'
> > timvf_worker.c:(.text+0x30c): undefined reference to `__atomic_store_2'
> >
> > Signed-off-by: Gavin Hu <gavin.hu@arm.com>
> > Reviewed-by: Honnappa Nagarahalli <Honnappa.Nagarahalli@arm.com>
>
>
>
> Following patch is part of upstream. Are you testing with following patch/upstream.
>
> With this patch from Nikhilesh, this __ atomic__ compiling issue was gone.
> The two patches fix the same issue.
> Should I abandon my patch?
The root cause is analyzed, the -latomic needs only when atomic
operations on variable are not aligned. The fix added to align variable where it was
not correct.This fix the performance issue as well, adding -latomic
will simply hide the problem by not letting to know the
alignment/performance issue. IMO, If there is no dependency with -latomic
in current code base then we should NOT add new dependency.
>
> I see this note on: https://clang.llvm.org/docs/Toolchain.html
> Note
> Clang does not currently automatically link against libatomic when using libgcc_s. You may need to manually add -latomic to support this configuration when using non-native atomic operations (if you see link errors referring to __atomic_* functions).
>
> commit 55fbc92d7800100628579643c9ee2770614fef10
> Author: Pavan Nikhilesh <pbhagavatula@caviumnetworks.com>
> Date: Wed May 9 02:56:00 2018 +0530
>
> event/octeontx: fix build with clang 6
>
> Clang 6 & 7 fail to naturally align packed structs due to this clang
> can't use 8byte atomic primitives and splits them into lesser atomic
> primitives. To use lesser atomic primitives we need to link libatomic
> (-latomic), instead supply alignment attribute to the compiler.
>
> timvf_worker.c:(.text+0x498): undefined reference to `__atomic_fetch_add_8'
> timvf_worker.c:(.text+0x525): undefined reference to `__atomic_store_2'
> timvf_worker.c:(.text+0x557): undefined reference to `__atomic_fetch_add_4'
> timvf_worker.c:(.text+0x5de): undefined reference to `__atomic_store_2'
>
> Fixes: f874c1eb1519 ("event/octeontx: create and free timer
> adapter")
>
> Reported-by: Andrew Rybchenko <arybchenko@solarflare.com>
> Signed-off-by: Pavan Nikhilesh <pbhagavatula@caviumnetworks.com>
> Acked-by: Jerin Jacob <jerin.jacob@caviumnetworks.com>
prev parent reply other threads:[~2018-05-16 5:05 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
[not found] <mailman.3990.1526375522.30695.dev@dpdk.org>
2018-05-16 2:08 ` Gavin Hu
2018-05-16 5:05 ` Jerin Jacob [this message]
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=20180516050505.GA16248@jerin \
--to=jerin.jacob@caviumnetworks.com \
--cc=Gavin.Hu@arm.com \
--cc=Jerin.JacobKollanukkaran@cavium.com \
--cc=dev@dpdk.org \
/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: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).