Add SocketCan J1939 constants and structs

[abalmos]

Add SocketCan J1939 constants and structs.

Blocking a PR to nix to wrap SocketCan's j1939 module.

[rust-highfive]

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

Please see the contribution instructions for more information.

[abalmos]

Looks like

enum {
	J1939_NLA_PAD,
	J1939_NLA_BYTES_ACKED,
	J1939_NLA_TOTAL_SIZE,
	J1939_NLA_PGN,
	J1939_NLA_SRC_NAME,
	J1939_NLA_DEST_NAME,
	J1939_NLA_SRC_ADDR,
	J1939_NLA_DEST_ADDR,
};

wasn't added to the kernel until 5.15.

I actually don't need these for my work. Should I just remove them from the PR or is there another way to deal with the CI being on an older kernel?

[Amanieu]

You can add an exception here: https://github.com/abalmos/libc/blob/a5470ce96f994347e14c5659442a7c086d2665c4/libc-test/build.rs#L3285

[abalmos]

@Amanieu Thanks! I think I pushed the fix. I am on 5.15, so the tests ran okay for me before this change

[abalmos]

@Amanieu I might need help with this error. It seems unrelated to this PR.

[Amanieu]

bad J1939_NO_ADDR value at byte 0: rust: 0 (0x0) != c 255 (0xff)

This means that the Rust value of the constant doesn't match the C value.

[abalmos]

Thanks @Amanieu! I'm not sure why libc-test worked on my machine from the get-go, but glad to see the CI passing now.

[Amanieu]

@bors r+

[bors]

📌 Commit 02042f0 has been approved by Amanieu

[bors]

⌛ Testing commit 02042f0 with merge 42232d8...

[bors]

💔 Test failed - checks-actions

[abalmos]

@Amanieu Re-added.

Sorry about that, I didn't realize there was more test environments then what was in GitHub actions.

[Amanieu]

@bors r+

[bors]

📌 Commit 99045cd has been approved by Amanieu

[bors]

⌛ Testing commit 99045cd with merge 99294b2...

[abalmos]

@Amanieu how do I check the error for this CI?

[Amanieu]

Click on the checks-actions link from bors. Ignore the semver tests, we allow failure for those.

[abalmos]

I had tried that but I don't have any meaningful output. I just see that it failed:

[Amanieu]

You need to use the bar on the left to select the actual build step that failed. Here is the failure:

  running: "musl-gcc" "-O0" "-ffunction-sections" "-fdata-sections" "-fPIC" "-g" "-fno-omit-frame-pointer" "-march=armv6" "-marm" "-mfpu=vfp" "-Wall" "-Wextra" "-Wall" "-Wextra" "-Werror" "-Wno-unused-parameter" "-Wno-type-limits" "-Wno-address-of-packed-member" "-Wno-unknown-warning-option" "-Wno-deprecated-declarations" "-D_GNU_SOURCE" "-D__GLIBC_USE_DEPRECATED_SCANF" "-o" "/checkout/target/arm-unknown-linux-musleabihf/debug/build/libc-test-87c193be97bba046/out/main.o" "-c" "/checkout/target/arm-unknown-linux-musleabihf/debug/build/libc-test-87c193be97bba046/out/main.c"
  cargo:warning=/checkout/target/arm-unknown-linux-musleabihf/debug/build/libc-test-87c193be97bba046/out/main.c:97:10: fatal error: linux/can/j1939.h: No such file or directory
  cargo:warning=   97 | #include <linux/can/j1939.h>
  cargo:warning=      |          ^~~~~~~~~~~~~~~~~~~
  cargo:warning=compilation terminated.
`

[abalmos]

I am stumped by this one ... it looks like the linux/can/j1939.h header is not present in the CI, yet it is running 20.04 and should have it based on Ubuntu repos. We develop on an arm board running Debian Buster without problem.

[abalmos]

@Amanieu Ah, I see. The test environment installs kernel headers cleaned for musl here:

libc/ci/install-musl.sh

Line 84 in 802f327

"https://github.com/sabotage-linux/kernel-headers/archive/v${KERNEL_HEADER_VER}.tar.gz" | \

but that project uses headers for kernel 4.19.88. j1939 was not added until v5.4.1.

I just pushed a commit that I believe will disable tests for musl. These are all kernel defines and types, so the testing done in the gnu environments ought to be enough. -- Please do let me know if you would prefer a different fix.

[Amanieu]

I think we should just go ahead and upgrade the version of the kernel headers used when testing musl to the latest available version.

[abalmos]

@Amanieu I would be happy to try to work on that, but the CI is using the latest set of headers available
here: https://github.com/sabotage-linux/kernel-headers

I don't know too much about musl and how all hooks with rust. Would you have any pointers on getting started with said upgrade?

[Amanieu]

Fair enough. In that case I think this is fine to land.

@bors r+

[bors]

📌 Commit 6991bfa has been approved by Amanieu

[bors]

⌛ Testing commit 6991bfa with merge 0ae4f07...

[bors]

💔 Test failed - checks-actions

[abalmos]

@Amanieu Is there any good way to run these tests locally?

[abalmos]

I accidentally put typedefs into skip_struct rather then skip_types.

I also found a comment that explains that testing kernel definitions in glibc targets only is sufficient, so I cleaned up the additions a bit by moving my skipped constant into that block.

Hopefully this version will pass :)

[Amanieu]

@bors r+

[bors]

📌 Commit 06a74e0 has been approved by Amanieu

[bors]

⌛ Testing commit 06a74e0 with merge bc9ea0b...

[bors]

☀️ Test successful - checks-actions, checks-cirrus-freebsd-11, checks-cirrus-freebsd-12, checks-cirrus-freebsd-13
Approved by: Amanieu
Pushing bc9ea0b to master...