From: "Stanisław Kardach" <kda@semihalf.com>
To: Tyler Retzlaff <roretzla@linux.microsoft.com>
Cc: "Morten Brørup" <mb@smartsharesystems.com>,
dev@dpdk.org, "Mattias Rönnblom" <mattias.ronnblom@ericsson.com>,
"Anatoly Burakov" <anatoly.burakov@intel.com>,
"Bruce Richardson" <bruce.richardson@intel.com>,
"David Christensen" <drc@linux.vnet.ibm.com>,
"Harry van Haaren" <harry.van.haaren@intel.com>,
"Konstantin Ananyev" <konstantin.v.ananyev@yandex.ru>,
"Min Zhou" <zhoumin@loongson.cn>,
"Ruifeng Wang" <ruifeng.wang@arm.com>
Subject: Re: [PATCH] eal: use C11 alignas instead of GCC attribute aligned
Date: Wed, 15 Nov 2023 23:43:05 +0100 [thread overview]
Message-ID: <CALVGJWJ1nT8WhVzqoHOHtzRwg3iS2QaETAfLJrw-dbKg+J5fDQ@mail.gmail.com> (raw)
In-Reply-To: <20231115210331.GA524@linuxonhyperv3.guj3yctzbm1etfxqx2vob5hsef.xx.internal.cloudapp.net>
On Wed, Nov 15, 2023 at 10:03 PM Tyler Retzlaff
<roretzla@linux.microsoft.com> wrote:
>
> On Wed, Nov 15, 2023 at 09:08:05PM +0100, Morten Brørup wrote:
> > > From: Tyler Retzlaff [mailto:roretzla@linux.microsoft.com]
> > > Sent: Wednesday, 15 November 2023 18.40
> > >
> > > Now that we have enabled C11 replace the use of __rte_cache_aligned
> > > and __rte_aligned(n) with alignas(RTE_CACHE_LINE_SIZE) and
> > > __rte_aligned(n) respectively.
> > >
> >
> > [...]
> >
> > > typedef union rte_xmm {
> > > + alignas(16)
> > > xmm_t x;
> > > uint8_t u8[XMM_SIZE / sizeof(uint8_t)];
> > > uint16_t u16[XMM_SIZE / sizeof(uint16_t)];
> > > uint32_t u32[XMM_SIZE / sizeof(uint32_t)];
> > > uint64_t u64[XMM_SIZE / sizeof(uint64_t)];
> > > double pd[XMM_SIZE / sizeof(double)];
> > > -} __rte_aligned(16) rte_xmm_t;
> > > +} rte_xmm_t;
> >
> > Your patch message should mention that C11 doesn't allow alignas() being applied to the declarations of struct/union types, so it is applied to the first field in the struct/union, which has the same effect.
>
> no problem, will add a note.
>
> >
> > Someone unfamiliar with alignas() would expect:
> >
> > -typedef union rte_xmm {
> > +typedef alignas(16) union rte_xmm {
> > [...]
> > -} __rte_aligned(16) rte_xmm_t;
> > +} rte_xmm_t;
> >
> > [...]
> >
> > > #ifndef RTE_VECT_RISCV_H
> > > #define RTE_VECT_RISCV_H
> > >
> > > +#include <stdalign.h>
> > > #include <stdint.h>
> > > #include "generic/rte_vect.h"
> > > #include "rte_common.h"
> > > @@ -23,13 +24,14 @@
> > > #define XMM_MASK (XMM_SIZE - 1)
> > >
> > > typedef union rte_xmm {
> > > + alignas(16) /* !! NOTE !! changed to 16 it looks like this was a
> > > bug? */
> > > xmm_t x;
> > > uint8_t u8[XMM_SIZE / sizeof(uint8_t)];
> > > uint16_t u16[XMM_SIZE / sizeof(uint16_t)];
> > > uint32_t u32[XMM_SIZE / sizeof(uint32_t)];
> > > uint64_t u64[XMM_SIZE / sizeof(uint64_t)];
> > > double pd[XMM_SIZE / sizeof(double)];
> > > -} __rte_aligned(8) rte_xmm_t;
> > > +} rte_xmm_t;
> >
> > Yes, this looks very much like a bug.
> > Even if a RISC-V CPU could handle alignment like that, it might interact with other software/hardware expecting type-sized alignment, i.e. 16-byte alignment, so partially using 8-byte alignment would cause bugs.
> >
> > It should be a separate patch with a Fixes tag.
>
> i'll submit a patch/fix for this so it is available and others can
> discuss if it should or shouldn't be merged for 23.11.
It is definitely a bug. Good catch. Since we did not have vector
extensions on our bring-up board, all xmm_t handling was essentially
scalar.
>
> >
> > We need to urgently decide if this bug should live on in DPDK 23.11, or if the fix should be included although we are very late in the release process.
> >
> > Stanislaw, what do you think?
> >
> > Furthermore, I wonder if it can be backported to stable, and to what extent backporting it would break the ABI/API.
> >
--
Best Regards,
Stanisław Kardach
next prev parent reply other threads:[~2023-11-15 22:43 UTC|newest]
Thread overview: 29+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-11-15 17:39 [PATCH] RFC: " Tyler Retzlaff
2023-11-15 17:39 ` [PATCH] eal: " Tyler Retzlaff
2023-11-15 18:13 ` Bruce Richardson
2023-11-15 18:27 ` Tyler Retzlaff
2023-11-15 20:08 ` Morten Brørup
2023-11-15 21:03 ` Tyler Retzlaff
2023-11-15 22:43 ` Stanisław Kardach [this message]
2023-11-16 10:12 ` Mattias Rönnblom
2024-01-25 18:37 ` [PATCH] RFC: " Tyler Retzlaff
2024-01-25 22:53 ` Morten Brørup
2024-01-25 23:31 ` Tyler Retzlaff
2024-01-26 10:05 ` Mattias Rönnblom
2024-01-26 10:18 ` Morten Brørup
2024-01-27 19:15 ` Mattias Rönnblom
2024-01-28 8:57 ` Morten Brørup
2024-01-28 10:00 ` Mattias Rönnblom
2024-01-29 19:43 ` Tyler Retzlaff
2024-01-30 8:08 ` Mattias Rönnblom
2024-01-30 17:39 ` Tyler Retzlaff
2024-01-30 17:59 ` Bruce Richardson
2024-01-30 18:01 ` Bruce Richardson
2024-01-30 18:04 ` Tyler Retzlaff
2024-01-30 18:18 ` Mattias Rönnblom
2024-01-31 16:04 ` Mattias Rönnblom
2024-01-30 8:09 ` Morten Brørup
2024-01-30 9:28 ` Mattias Rönnblom
2024-01-30 10:17 ` Morten Brørup
2024-01-30 13:00 ` Morten Brørup
2024-01-30 17:54 ` Tyler Retzlaff
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=CALVGJWJ1nT8WhVzqoHOHtzRwg3iS2QaETAfLJrw-dbKg+J5fDQ@mail.gmail.com \
--to=kda@semihalf.com \
--cc=anatoly.burakov@intel.com \
--cc=bruce.richardson@intel.com \
--cc=dev@dpdk.org \
--cc=drc@linux.vnet.ibm.com \
--cc=harry.van.haaren@intel.com \
--cc=konstantin.v.ananyev@yandex.ru \
--cc=mattias.ronnblom@ericsson.com \
--cc=mb@smartsharesystems.com \
--cc=roretzla@linux.microsoft.com \
--cc=ruifeng.wang@arm.com \
--cc=zhoumin@loongson.cn \
/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).