From: Adhemerval Zanella <adhemerval.zanella@linaro.org>
To: Carlos O'Donell <carlos@redhat.com>, libc-alpha@sourceware.org
Subject: Re: [PATCH v2 14/25] y2038: Use a common definition for msqid_ds
Date: Mon, 7 Jun 2021 15:29:49 -0300 [thread overview]
Message-ID: <d44eda35-16d2-0f51-0cda-e8c6ad860298@linaro.org> (raw)
In-Reply-To: <c8ff63aa-8a47-05a8-0228-66596d31dff9@redhat.com>
On 04/06/2021 16:38, Carlos O'Donell wrote:
> On 5/18/21 4:56 PM, Adhemerval Zanella wrote:
>> From: Lukasz Majewski <lukma@denx.de>
>>
>> Instead of replicate the same definitions from struct_msqid64_ds.h
>> on the multiple struct_msqid_ds.h, use a common header which is included
>> when required (struct_msqid64_ds_helper.h).
>>
>> The __USE_TIME_BITS64 is not defined internally yet, although the
>> internal header is used when building the 64-bit stat implementations.
>
> Requesting a v3 please.
>
> Please add the __glibc_reserved* members to the structures to preserve
> maximum compatibility with the kernel structure.
>
> What I would like to see is that __glibc_reserved4 and __glibc_reserved5
> are *preserved* for both ABIs in the dual-time case. The kernel for msqid
> still has the following:
>
> 26 struct msqid64_ds {
> 27 struct ipc64_perm msg_perm;
> 28 #if __BITS_PER_LONG == 64
> 29 long msg_stime; /* last msgsnd time */
> 30 long msg_rtime; /* last msgrcv time */
> 31 long msg_ctime; /* last change time */
> 32 #else
> 33 unsigned long msg_stime; /* last msgsnd time */
> 34 unsigned long msg_stime_high;
> 35 unsigned long msg_rtime; /* last msgrcv time */
> 36 unsigned long msg_rtime_high;
> 37 unsigned long msg_ctime; /* last change time */
> 38 unsigned long msg_ctime_high;
> 39 #endif
> 40 unsigned long msg_cbytes; /* current number of bytes on queue */
> 41 unsigned long msg_qnum; /* number of messages in queue */
> 42 unsigned long msg_qbytes; /* max number of bytes on queue */
> 43 __kernel_pid_t msg_lspid; /* pid of last msgsnd */
> 44 __kernel_pid_t msg_lrpid; /* last receive pid */
> 45 unsigned long __unused4;
> 46 unsigned long __unused5;
> 47 };
>
> For maximum compatibility with future kernel changes we should keep
> the unused fields in the glibc equivalent structure. If the values are
> ever used then they would already be allocated as part of the size of
> the structure without the need for an ABI change. Given that these are
> somewhat legacy APIs we should be aiming for preserving existing behaviour
> rather than aiming for the smallest possible size structure.
>
> On i686 the new 64-bit time_t struct is only 80 bytes, while the original
> struct size is 88 bytes. Adding the reserved entries brings this back to
> 88 bytes in the new ABI. We can add static asserts if we think we need to
> check exactly for alignment of 64-bit time_t, but since we copy from a
> kernel structure I'm largely worried about the ABI being too small to cary
> the information that we might need to copy from the matching kernel struct.
>
Alright, I already sent my findings that it would increase the msqid_ds
size for _TIME_BITS=64 on some legacy ABIs (arm, mips, s390) but as
we discussed on weekly call it should be fine.
I don't think we need a static assert to check for alignment here (since
glibc export struct for _TIME_BITS might not be potentially equal to
kernel expected one and we already do the heavy lifting in msgctl to
copy and adjust the fields).
>
> This is a spurious cleanup from the last patch IMO. If possible please move it
> into the previous commit.
>
> s/defitions/definitions/g
>
>> Copyright (C) 2021 Free Software Foundation, Inc.
>> This file is part of the GNU C Library.
>>
Ack, I just removed this change.
>> diff --git a/sysdeps/unix/sysv/linux/bits/types/struct_msqid64_ds.h b/sysdeps/unix/sysv/linux/bits/types/struct_msqid64_ds.h
>> index 43e8cd7cfc..992734914a 100644
>> --- a/sysdeps/unix/sysv/linux/bits/types/struct_msqid64_ds.h
>> +++ b/sysdeps/unix/sysv/linux/bits/types/struct_msqid64_ds.h
>> @@ -25,14 +25,6 @@
>> #else
>> struct __msqid64_ds
>> {
>> - struct ipc_perm msg_perm; /* structure describing operation permission */
>> - __time64_t msg_stime; /* time of last msgsnd command */
>> - __time64_t msg_rtime; /* time of last msgsnd command */
>> - __time64_t msg_ctime; /* time of last change */
>> - __syscall_ulong_t __msg_cbytes; /* current number of bytes on queue */
>> - msgqnum_t msg_qnum; /* number of messages currently on queue */
>> - msglen_t msg_qbytes; /* max number of bytes allowed on queue */
>> - __pid_t msg_lspid; /* pid of last msgsnd() */
>> - __pid_t msg_lrpid; /* pid of last msgrcv() */
>> +# include <bits/types/struct_msqid64_ds_helper.h>
>
> OK. Used unconditionally since this is 64-bit time_t.
>
>> };
>> #endif
>> diff --git a/sysdeps/unix/sysv/linux/bits/types/struct_msqid64_ds_helper.h b/sysdeps/unix/sysv/linux/bits/types/struct_msqid64_ds_helper.h
>> new file mode 100644
>> index 0000000000..02dfddaa2b
>> --- /dev/null
>> +++ b/sysdeps/unix/sysv/linux/bits/types/struct_msqid64_ds_helper.h
>> @@ -0,0 +1,28 @@
>> +/* Common defintions for struct msqid_ds with 64 bit time.
>
> s/defintions/definitions/g
Ack.
>
>> + Copyright (C) 2021 Free Software Foundation, Inc.
>> + This file is part of the GNU C Library.
>> +
>> + The GNU C Library is free software; you can redistribute it and/or
>> + modify it under the terms of the GNU Lesser General Public
>> + License as published by the Free Software Foundation; either
>> + version 2.1 of the License, or (at your option) any later version.
>> +
>> + The GNU C Library is distributed in the hope that it will be useful,
>> + but WITHOUT ANY WARRANTY; without even the implied warranty of
>> + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
>> + Lesser General Public License for more details.
>> +
>> + You should have received a copy of the GNU Lesser General Public
>> + License along with the GNU C Library. If not, see
>> + <https://www.gnu.org/licenses/>. */
>> +
>
> Reviewing...
>
>> + /* Content of internal __msqid64_ds. */
>> + struct ipc_perm msg_perm; /* structure describing operation permission */
>
> OK.
>
>> + __time64_t msg_stime; /* time of last msgsnd command */
>> + __time64_t msg_rtime; /* time of last msgsnd command */
>> + __time64_t msg_ctime; /* time of last change */
>
> OK. Usually a tuple of * and *_high (endian swapping for order applied).
>
>> + __syscall_ulong_t __msg_cbytes; /* current number of bytes on queue */
>
> OK.
>
>> + msgqnum_t msg_qnum; /* number of messages currently on queue */
>
> OK.
>
>> + msglen_t msg_qbytes; /* max number of bytes allowed on queue */
>
> OK.
>
>> + __pid_t msg_lspid; /* pid of last msgsnd() */
>> + __pid_t msg_lrpid; /* pid of last msgrcv() */
>
> OK.
>
> Add __glibc_reserved4/5 entries here.
>
Ack.
next prev parent reply other threads:[~2021-06-07 18:29 UTC|newest]
Thread overview: 92+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-05-18 20:55 [PATCH v2 00/25] Add 64 bit time support on legacy ABIs Adhemerval Zanella
2021-05-18 20:55 ` [PATCH v2 01/25] linux: mips: Split libpthread.abilist in n32 and n64 Adhemerval Zanella
2021-05-19 8:24 ` Lukasz Majewski
2021-05-20 6:38 ` Florian Weimer
2021-05-20 10:43 ` Adhemerval Zanella
2021-06-04 19:29 ` Carlos O'Donell
2021-05-18 20:55 ` [PATCH v2 02/25] linux: mips: Split librt.abilist " Adhemerval Zanella
2021-05-19 8:25 ` Lukasz Majewski
2021-06-04 19:29 ` Carlos O'Donell
2021-05-18 20:55 ` [PATCH v2 03/25] linux: mips: Split libanl.abilist " Adhemerval Zanella
2021-05-19 8:25 ` Lukasz Majewski
2021-06-04 19:30 ` Carlos O'Donell
2021-05-18 20:55 ` [PATCH v2 04/25] linux: s390: Add libanl.abilist in s390 and s390x Adhemerval Zanella
2021-05-19 8:26 ` Lukasz Majewski
2021-06-04 19:30 ` Carlos O'Donell
2021-05-18 20:55 ` [PATCH v2 05/25] linux: Add fallback for 64-bit time_t SO_{RCV, SND}TIMEO Adhemerval Zanella
2021-05-19 8:36 ` [PATCH v2 05/25] linux: Add fallback for 64-bit time_t SO_{RCV,SND}TIMEO Lukasz Majewski
2021-05-20 6:44 ` [PATCH v2 05/25] linux: Add fallback for 64-bit time_t SO_{RCV, SND}TIMEO Florian Weimer
2021-05-20 18:01 ` Adhemerval Zanella
2021-05-21 18:37 ` Florian Weimer
2021-05-21 19:17 ` Adhemerval Zanella
2021-06-04 19:30 ` [PATCH v2 05/25] linux: Add fallback for 64-bit time_t SO_{RCV,SND}TIMEO Carlos O'Donell
2021-06-07 17:52 ` Adhemerval Zanella
2021-05-18 20:55 ` [PATCH v2 06/25] linux: Add fallback for 64-bit time_t SO_TIMESTAMP{NS} Adhemerval Zanella
2021-05-19 8:50 ` Lukasz Majewski
2021-05-20 6:50 ` Florian Weimer
2021-05-20 18:46 ` Adhemerval Zanella
2021-05-21 18:38 ` Florian Weimer
2021-05-21 19:02 ` Adhemerval Zanella
2021-06-04 19:30 ` Carlos O'Donell
2021-05-18 20:55 ` [PATCH v2 07/25] linux: Add recvvmsg " Adhemerval Zanella
2021-05-19 9:02 ` Lukasz Majewski
2021-06-04 19:30 ` Carlos O'Donell
2021-05-18 20:55 ` [PATCH v2 08/25] y2038: Add __USE_TIME_BITS64 support for time_t Adhemerval Zanella
2021-05-19 9:02 ` Lukasz Majewski
2021-06-04 19:30 ` Carlos O'Donell
2021-05-18 20:55 ` [PATCH v2 09/25] y2038: Add __USE_TIME_BITS64 support for struct timeval Adhemerval Zanella
2021-05-19 9:03 ` Lukasz Majewski
2021-06-04 19:31 ` Carlos O'Donell
2021-05-18 20:55 ` [PATCH v2 10/25] y2038: Add __USE_TIME_BITS64 support for struct timespec Adhemerval Zanella
2021-05-19 9:03 ` Lukasz Majewski
2021-06-04 19:31 ` Carlos O'Donell
2021-05-18 20:55 ` [PATCH v2 11/25] y2038: Add __USE_TIME_BITS64 support for struct utimbuf Adhemerval Zanella
2021-05-19 9:04 ` Lukasz Majewski
2021-06-04 19:31 ` Carlos O'Donell
2021-05-18 20:56 ` [PATCH v2 12/25] y2038: linux: Add __USE_TIME_BITS64 support for struct timex Adhemerval Zanella
2021-05-19 9:04 ` Lukasz Majewski
2021-06-04 19:31 ` Carlos O'Donell
2021-05-18 20:56 ` [PATCH v2 13/25] y2038: Use a common definition for stat Adhemerval Zanella
2021-06-04 19:37 ` Carlos O'Donell
2021-06-07 18:07 ` Adhemerval Zanella
2021-05-18 20:56 ` [PATCH v2 14/25] y2038: Use a common definition for msqid_ds Adhemerval Zanella
2021-06-04 19:38 ` Carlos O'Donell
2021-06-07 18:29 ` Adhemerval Zanella [this message]
2021-05-18 20:56 ` [PATCH v2 15/25] y2038: Use a common definition for semid_ds Adhemerval Zanella
2021-05-19 9:09 ` Lukasz Majewski
2021-06-04 19:38 ` Carlos O'Donell
2021-06-07 18:46 ` Adhemerval Zanella
2021-05-18 20:56 ` [PATCH v2 16/25] y2038: Use a common definition for shmid_ds Adhemerval Zanella
2021-05-19 9:09 ` Lukasz Majewski
2021-06-04 19:38 ` Carlos O'Donell
2021-05-18 20:56 ` [PATCH v2 17/25] y2038: Add __USE_TIME_BITS64 support for socket-constants.h Adhemerval Zanella
2021-05-19 9:13 ` Lukasz Majewski
2021-06-04 19:38 ` Carlos O'Donell
2021-05-18 20:56 ` [PATCH v2 18/25] time: Add 64 bit time support for getdate Adhemerval Zanella
2021-05-19 9:15 ` Lukasz Majewski
2021-06-04 19:38 ` Carlos O'Donell
2021-05-18 20:56 ` [PATCH v2 19/25] y2038: Add support for 64 bit time on legacy ABIs Adhemerval Zanella
2021-05-19 9:18 ` Lukasz Majewski
2021-05-20 6:58 ` Florian Weimer
2021-05-20 10:37 ` Adhemerval Zanella
2021-06-04 19:38 ` Carlos O'Donell
2021-05-18 20:56 ` [PATCH v2 20/25] posix: Add glob64 with 64 bit time_t support Adhemerval Zanella
2021-05-19 10:44 ` Lukasz Majewski
2021-06-04 19:39 ` Carlos O'Donell
2021-06-07 18:52 ` Adhemerval Zanella
2021-05-18 20:56 ` [PATCH v2 21/25] io: Add fts64 " Adhemerval Zanella
2021-05-19 10:50 ` Lukasz Majewski
2021-06-04 19:39 ` Carlos O'Donell
2021-05-18 20:56 ` [PATCH v2 22/25] io: Add ftw64 " Adhemerval Zanella
2021-05-19 10:57 ` Lukasz Majewski
2021-06-04 19:39 ` Carlos O'Donell
2021-05-18 20:56 ` [PATCH v2 23/25] libsupport: Add 64 bit time_t support for time functions Adhemerval Zanella
2021-05-19 11:00 ` Lukasz Majewski
2021-06-04 19:39 ` Carlos O'Donell
2021-05-18 20:56 ` [PATCH v2 24/25] libsupport: Add 64 bit time_t support for stat functions Adhemerval Zanella
2021-05-19 11:04 ` Lukasz Majewski
2021-06-04 19:39 ` Carlos O'Donell
2021-05-18 20:56 ` [PATCH v2 25/25] y2038: Add test coverage Adhemerval Zanella
2021-05-19 11:08 ` Lukasz Majewski
2021-06-04 19:39 ` Carlos O'Donell
2021-06-04 19:29 ` [PATCH v2 00/25] Add 64 bit time support on legacy ABIs Carlos O'Donell
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=d44eda35-16d2-0f51-0cda-e8c6ad860298@linaro.org \
--to=adhemerval.zanella@linaro.org \
--cc=carlos@redhat.com \
--cc=libc-alpha@sourceware.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 read-only IMAP folder(s) and NNTP newsgroup(s).