public inbox for libc-alpha@sourceware.org
 help / color / mirror / Atom feed
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.

  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).