Switch to 64 bit file and time APIs on GNU libc for 32bit systems #3175

snogge · 2023-03-29T12:17:40Z

Use the 64 bit types and APIs included in GNU libc also for 32-bit systems. These are the types and APIs used when you compile your C code against GNU libc headers with the preprocessor options -D_FILE_OFFSET_BITS=64 -D_TIME_BITS=64.

This is required for 2038 compatible software built for 32-bit systems.

This PR is not done, I need some guidance with finishing it.

Does it need some sort of configuration options to turn this on and off?
There is a conflict for some of the stat functions where two signatures link against the same symbol. For C this is not a problem, but we probably need to make struct stat and struct stat64 be different names for the same struct. I don't know how to do that.
It's possible that some types should be moved/split/joined in the file hierarchy. I'd appreciate some feedback on that, but I'm waiting on the answer to question 1 before moving forward.

I've done one commit for each modified type so it is easier to review the changes in isolation.

rustbot · 2023-03-29T12:17:45Z

Thanks for the pull request, and welcome! The Rust team is excited to review your changes, and you should hear from @JohnTitor (or someone else) soon.

Please see the contribution instructions for more information. Namely, in order to ensure the minimum review times lag, PR authors and assigned reviewers should ensure that the review label (S-waiting-on-review and S-waiting-on-author) stays updated, invoking these commands when appropriate:

@rustbot author: the review is finished, PR author should check the comments and take action accordingly
@rustbot review: the author is ready for a review, this PR will be queued again in the reviewer's queue

snogge · 2023-03-29T13:24:08Z

Oops, left some debug/test commits in there. Removed now.

snogge · 2023-04-11T10:25:46Z

Hi @JohnTitor , do you have any feedback on this change?
Is there anything I can do to facilitate the review?
I'd appreciate any hints on solving question 2:

There is a conflict for some of the stat functions where two signatures link against the same symbol. For C this is not a problem, but we probably need to make struct stat and struct stat64 be different names for the same struct. I don't know how to do that.

I'm not a Rust developer and not sure how I can create that type alias.

bors · 2023-04-23T02:57:43Z

☔ The latest upstream changes (presumably #3218) made this pull request unmergeable. Please resolve the merge conflicts.

JohnTitor · 2023-04-24T17:40:21Z

Thanks for the PR! But I think this is a huge breaking change and we have to discuss it before reviewing the changes (or, even submitting a PR). Could you open an issue instead?

snogge · 2023-04-25T07:39:29Z

Sure, I'll do that.

snogge · 2023-04-25T08:03:44Z

I'm working on updating this PR to handle problem 2 mentioned above and to pass tests on more platforms. Expect an update later this week.

snogge · 2023-04-25T12:04:04Z

Sorry for the many pushes, that was meant only for my fork. Anyway, now the PR should handle the duplicate functions mentioned in problem 2, and also pass tests on more platforms.

bors · 2023-05-06T06:29:22Z

☔ The latest upstream changes (presumably #3237) made this pull request unmergeable. Please resolve the merge conflicts.

snogge · 2023-09-07T13:05:38Z

Just rebased on main. Passes all tests in main.yml and bors.yml workflows.
Still not very clean though, feedback on code style would be welcome.

I considered adding a cargo feature to control this, but as that would be braking the ABI I guess I shouldn't?

bors · 2023-11-12T11:53:17Z

☔ The latest upstream changes (presumably #3437) made this pull request unmergeable. Please resolve the merge conflicts.

Set the basic types correctly for gnu_time64_abi (_TIME_BITS=64 and _FILE_OFFSET_BITS=64).

Set the link names of relevant symbols to use be the same as when a C program is built against GNU libc with -D_TIME_BITS=64 and -D_FILE_OFFSET_BITS=64.

The actual values may be different on 32bit archs and glibc with 64-bit time.

Also add the F_[SG]ET*64 constants.

For 64 bit time on 32 bit linux glibc timeval.tv_usec is actually __suseconds64_t (64 bits) while suseconds_t is still 32 bits.

Big-endian platforms wants 32 bits of padding before tv_nsec, little-endian after. GNU libc always uses long for tv_nsec.

Do not use gnu/b32/time*.rs for riscv32 and sparc. Add timex to gnu/b32/(riscv32|sparc)/mod.rs instead.

Move the stat struct from gnu/b32/mod.rs to gnu/b32/time*.rs For arm, mips, powerpc, riscv32, and x86, move stat64 to time32.rs and add an stat64 = stat alias to time64.rs. For sparc, do the same for statfs64 and statvfs64.

In Linux Tier1, run the tests for i686-unknown-linux-gnu with RUST_LIBC_TIME_BITS set to 32, 64, and default. In Linux Tier2, run the tests for arm-unknown-linux-gnueabihf and powerpc-unknown-linux-gnu with RUST_LIBC_TIME_BITS set to 32, 64, and default. Use RUST_LIBC_TIME_BITS=defaults for the other platforms. In Build Channels Linux, build the relevant platforms with RUST_LIBC_TIME_BITS unset and set to 32 and 64.

snogge · 2024-08-15T08:53:30Z

@JohnTitor @tgross35 @joshtriplett
I apologize for taking this shotgun-like approach for attention.
This PR is now ~1.5 years old, and I'm still not sure whether there is any plan to include it in the 1.0 release and whether my approach is acceptable.
I realize it is a large and controversial change, but I'd really like some feedback on the PRs viability and the choices made.

tgross35 · 2024-08-29T08:12:15Z

@snogge thank you for the change and for being willing to keep up with it. We need this change and I think something like this may be about the only reasonable approach.

I looked at this and have some small feedback, but I need to double check with the libs team that this is the path we want/need to go. I'm going to try to get a meeting scheduled for this which will probably take about a month, after that we will know what is needed to get this merged. Hold off on any rebases until then just in case, but I think we will still want most of this even if something like the config has to change.

snogge · 2024-10-21T07:41:37Z

@tgross35 Any updates? It's been almost two months now.

snogge · 2024-11-19T08:47:16Z

Ping @tgross35 , it's been another month and rebasing will just be getting more and more difficult.

tgross35 · 2024-11-19T09:20:40Z

Timing, I was actually just looking at this :) I think we just need to start merging small parts of this to unblock things. Would it be possible to split the following changes to a new PR?

The changes to src/unix/linux_like/linux/mod.rs
Adding gnu_time64_abi to ALLOWED_CFGS in build.rs

If we can get that merged as a basis then the rest (testing, platform support, defaults) can fall into place as we figure it out.

tgross35 · 2024-11-20T04:44:10Z

Actually, a more general request than that. Could uapi-related changes be split to a PR that we can merge first?

I mentioned this in the musl PR at #3791 (comment) but we should have a config linux_time_bits64 that corresponds to __USE_TIME_BITS64 used by uapi (example in struct input_event, my suggested rust field implementation). This should be used for anything that isn't glibc-specific.

Doing this first makes the rest of the changes here a lot smaller and unblocks the musl PR as well.

tgross35 · 2024-11-23T00:37:18Z

Per the above, I think getting the Linux cfg in place is a reasonable first separate first since it affects all libraries and is a smaller portion of this change. After that is in place it will be easier to merge the rest.

Let me know if you need any help here.

@rustbot author

snogge · 2024-11-25T13:42:59Z

I'm working on a new PR, but it's been a while so I need to double check some things.

snogge · 2024-11-25T23:40:58Z

@tgross35 new PR for linux_time_bits64: #4148

rustbot assigned JohnTitor Mar 29, 2023

rustbot added the S-waiting-on-review label Mar 29, 2023

snogge force-pushed the time_bits=64 branch from 78cec5c to c22d06d Compare March 29, 2023 13:23

JohnTitor added breakage-candidate S-blocked I-needs-decision and removed S-waiting-on-review labels Apr 1, 2023

snogge force-pushed the time_bits=64 branch 7 times, most recently from 5dfc8ed to 7d2c859 Compare April 24, 2023 13:36

snogge force-pushed the time_bits=64 branch from 7d2c859 to 6e00e1b Compare April 25, 2023 06:41

snogge mentioned this pull request Apr 25, 2023

Use the glibc 64bit time APIs to mitigate the Y2038 problem on 32bit platforms #3223

Open

snogge force-pushed the time_bits=64 branch 2 times, most recently from 5314484 to eb997cb Compare April 25, 2023 12:02

snogge force-pushed the time_bits=64 branch from eb997cb to 4d16795 Compare September 7, 2023 13:03

snogge added 17 commits August 15, 2024 09:23

gnu: Handle basic file and time types for 32bit systems with time64

6b82eb2

Set the basic types correctly for gnu_time64_abi (_TIME_BITS=64 and _FILE_OFFSET_BITS=64).

gnu: Use _TIME_BITS and _FILE_OFFSET_BITS=64 versions of glibc symbols

107177a

Set the link names of relevant symbols to use be the same as when a C program is built against GNU libc with -D_TIME_BITS=64 and -D_FILE_OFFSET_BITS=64.

linux: Set SO_TIMESTAMP* and SO_RCVTIMEO and SO_SNDTIMEO

5a0eb4b

The actual values may be different on 32bit archs and glibc with 64-bit time.

linux: Set RLIM_INFINITY for mips GNU libc with _FILE_OFFSET_BITS=64

9c46b6c

gnu: Update F_GETLK and F_GETLK64 for 64-bit time

ecf680c

gnu: Update F_SETLK and F_SETLKW for 64-bit time

8f514a1

Also add the F_[SG]ET*64 constants.

linux: Update struct input_event for glibc _TIME_BITS=64

5213d0b

gnu: Handle timeval.tv_usec for glibc 64-bit time_t

50a931d

For 64 bit time on 32 bit linux glibc timeval.tv_usec is actually __suseconds64_t (64 bits) while suseconds_t is still 32 bits.

gnu 32 powerpc: Rename powerpc.rs -> powerpc/mod.rs

c4a27a8

gnu: Update struct shmid_ds for 64-bit time

d2af68c

gnu: Update struct msqid_ds for 64-bit time

bc553de

gnu: Update struct semid_ds for 64-bit time

f2ff593

gnu: Update struct aiocb for _TIME_BITS=64

83c60cd

gnu: Update struct timespec for _TIME_BITS=64

49d1137

Big-endian platforms wants 32 bits of padding before tv_nsec, little-endian after. GNU libc always uses long for tv_nsec.

gnu: Move struct timex to gnu/b64 and gnu/b32/time*.rs

fefc83c

Do not use gnu/b32/time*.rs for riscv32 and sparc. Add timex to gnu/b32/(riscv32|sparc)/mod.rs instead.

gnu: Update the stat* structs for 64-bit time

d9a25a7

Move the stat struct from gnu/b32/mod.rs to gnu/b32/time*.rs For arm, mips, powerpc, riscv32, and x86, move stat64 to time32.rs and add an stat64 = stat alias to time64.rs. For sparc, do the same for statfs64 and statvfs64.

snogge force-pushed the time_bits=64 branch from 473fcae to 1db367e Compare August 15, 2024 08:25

rustbot added S-waiting-on-author and removed S-waiting-on-review labels Nov 23, 2024

snogge mentioned this pull request Nov 26, 2024

Add new cfg linux_time_bits64 corresponding to __USE_TIME_BITS64 #4148

Open

3 tasks

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Switch to 64 bit file and time APIs on GNU libc for 32bit systems #3175

Switch to 64 bit file and time APIs on GNU libc for 32bit systems #3175

snogge commented Mar 29, 2023

rustbot commented Mar 29, 2023

snogge commented Mar 29, 2023

snogge commented Apr 11, 2023

bors commented Apr 23, 2023

JohnTitor commented Apr 24, 2023 •

edited

Loading

snogge commented Apr 25, 2023

snogge commented Apr 25, 2023

snogge commented Apr 25, 2023

bors commented May 6, 2023

snogge commented Sep 7, 2023

bors commented Nov 12, 2023

snogge commented Aug 15, 2024

tgross35 commented Aug 29, 2024 •

edited

Loading

snogge commented Oct 21, 2024

snogge commented Nov 19, 2024

tgross35 commented Nov 19, 2024

tgross35 commented Nov 20, 2024

tgross35 commented Nov 23, 2024

snogge commented Nov 25, 2024

snogge commented Nov 25, 2024

Switch to 64 bit file and time APIs on GNU libc for 32bit systems #3175

Are you sure you want to change the base?

Switch to 64 bit file and time APIs on GNU libc for 32bit systems #3175

Conversation

snogge commented Mar 29, 2023

rustbot commented Mar 29, 2023

snogge commented Mar 29, 2023

snogge commented Apr 11, 2023

bors commented Apr 23, 2023

JohnTitor commented Apr 24, 2023 • edited Loading

snogge commented Apr 25, 2023

snogge commented Apr 25, 2023

snogge commented Apr 25, 2023

bors commented May 6, 2023

snogge commented Sep 7, 2023

bors commented Nov 12, 2023

snogge commented Aug 15, 2024

tgross35 commented Aug 29, 2024 • edited Loading

snogge commented Oct 21, 2024

snogge commented Nov 19, 2024

tgross35 commented Nov 19, 2024

tgross35 commented Nov 20, 2024

tgross35 commented Nov 23, 2024

snogge commented Nov 25, 2024

snogge commented Nov 25, 2024

JohnTitor commented Apr 24, 2023 •

edited

Loading

tgross35 commented Aug 29, 2024 •

edited

Loading