Don't adopt systemd boot loaders #462

dbnicholson · 2023-04-11T16:58:47Z

Skip adoption if systemd-boot or systemd-stub are used since bootctl will handle updating them. The 2nd commit is IMO correct but more controversial. I needed it to test that our sd-boot system would be detected and skipped. It did work in that case, though:

Apr 11 10:56:49 endless bootupd[3086]: [DEBUG bootupd::efi] Reusing existing "/boot"
Apr 11 10:56:49 endless bootupd[3086]: [TRACE bootupd::efi] No CoreOS aleph detected
Apr 11 10:56:49 endless bootupd[3086]: [TRACE bootupd::efi] Skipping adoption for "systemd-boot 251.2+dev56.9738984-22bem1"

I'm not that familiar with writing rust tests, but I think the UTF-16 string function could be written to take an optional root and then a test that writes various stuff into <testroot>/sys/firmware/efi/efivars/<somevar> could be added.

openshift-ci · 2023-04-11T16:58:59Z

Hi @dbnicholson. Thanks for your PR.

I'm waiting for a coreos member to verify that this patch is reasonable to test. If it is, they should reply with /ok-to-test on its own line. Until that is done, I will not automatically test new commits in this PR, but the usual testing commands by org members will still work. Regular contributors should join the org to skip this step.

Once the patch is verified, the new status will be reflected by the ok-to-test label.

I understand the commands that are listed here.

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.

cgwalters · 2023-05-04T21:09:35Z

/ok-to-test

cgwalters

Looks sane overall, just a few minor nits

cgwalters · 2023-05-04T21:11:25Z

src/efi.rs

+        return None;
+    }
+    let path = efivars.join(name);
+    if !path.exists() {


Optional cleanup in the future: We do have an open_file_optional helper

I didn't notice that.

cgwalters · 2023-05-05T01:09:56Z

src/efi.rs

+    // Drop the last byte if there's an odd number.
+    let size = slice.len() / 2;
+    let v: Vec<u16> = (0..size)
+        .map(|i| u16::from_ne_bytes([slice[2 * i], slice[2 * i + 1]]))


I think we can use https://doc.rust-lang.org/std/primitive.slice.html#method.chunks here

Oh, that does look nicer. My rust noobness shows through.

Your code is great honestly! Learning a language (programming or not) is always a journey; even the relatively small std in Rust has a lot of surface area and idioms.

cgwalters · 2023-05-05T01:11:26Z

src/efi.rs

@@ -105,6 +110,70 @@ impl Efi {
    }
 }

+/// Convert a nul-terminated UTF-16 byte array to a String.
+fn string_from_utf16_bytes(slice: &[u8]) -> String {
+    // For some reason, systemd appends 3 nul bytes after the string.


Hmm, messy. But I wonder if we should care about this? Maybe instead we return a Vec<u8> since in the end all we care about is if it starts with the ASCII string systemd right?

OK as is though.

You can't just send an array of bytes (which you already have from reading the file contents) back since they're be misinterpreted as UTF-8 since every other byte is nul. That's why you have to convert it to Vec<u16> and parse it as UTF-16. I don't believe there's any spec that says you have to write UTF-16 strings to EFI variables. I think that's just a convention systemd follows since that's how other vendors do it. I think I read a comment in the systemd code that they append 3 nul bytes so that in case they start parsing in the middle of the buffer they'll still end up with a valid C string. However, stripping an odd numbered trailing byte seems like good practice.

When I wrote roughly the same code in Python where I didn't have to change types, I walked backwards from the end of the buffer until I found no more pairs of nul bytes. Here I let U16CString handle that part.

Yes...hm. What about going the other way, and converting from systemd in UTF-8 to UTF-16, and then using https://doc.rust-lang.org/std/primitive.slice.html#method.starts_with ?

That way we can also just ignore the trailing NULs.

cgwalters

I think this is fine as is, we can do followups later.

cgwalters · 2023-05-05T12:48:26Z

Hmm...I don't think the CI failure here is related, I see it on other PRs, but now I'm a bit confused why CI worked in #450 but not elsewhere.

If the system is booted with systemd-boot or systemd-stub then `bootctl` will handle updating the boot components. Both of these components write identifying information to EFI variables as NUL terminated UTF-16 strings. Check these variables to detect if either systemd-boot or systemd-stub are in use and skip adoption if so.

Some systems use the ESP as the entire `/boot` filesystem. This check needs to go before `boot/efi` since that would also detect the FAT filesystem but would expect the `EFI` directory at `/boot/efi/EFI` instead of `/boot/EFI`. A more typical system with the ESP mounted at `/boot/efi` would not have use FAT for the `/boot` filesystem and would be skipped in this check.

cgwalters · 2023-05-05T13:50:48Z

Tried rebasing this just to rerun CI and see if the failure is somehow fixed in main but I think the problem is rooted in #229

openshift-ci bot added the needs-ok-to-test label Apr 11, 2023

dbnicholson force-pushed the no-adopt-systemd-boot branch from c7dd388 to f4f0ce6 Compare April 11, 2023 17:03

openshift-ci bot added ok-to-test and removed needs-ok-to-test labels May 4, 2023

cgwalters approved these changes May 5, 2023

View reviewed changes

cgwalters added the triaged This issue was triaged label May 5, 2023

cgwalters approved these changes May 5, 2023

View reviewed changes

dbnicholson added 2 commits May 5, 2023 09:49

cgwalters force-pushed the no-adopt-systemd-boot branch from f4f0ce6 to 8c5d4e7 Compare May 5, 2023 13:50

cgwalters merged commit 9d80017 into coreos:main May 5, 2023

dbnicholson deleted the no-adopt-systemd-boot branch May 8, 2023 14:56

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Don't adopt systemd boot loaders #462

Don't adopt systemd boot loaders #462

dbnicholson commented Apr 11, 2023

openshift-ci bot commented Apr 11, 2023

cgwalters commented May 4, 2023

cgwalters left a comment

cgwalters May 4, 2023

dbnicholson May 8, 2023

cgwalters May 5, 2023

dbnicholson May 8, 2023

cgwalters May 8, 2023

cgwalters May 5, 2023

dbnicholson May 8, 2023

cgwalters May 8, 2023 •

edited

Loading

cgwalters left a comment

cgwalters commented May 5, 2023

cgwalters commented May 5, 2023

Don't adopt systemd boot loaders #462

Don't adopt systemd boot loaders #462

Conversation

dbnicholson commented Apr 11, 2023

openshift-ci bot commented Apr 11, 2023

cgwalters commented May 4, 2023

cgwalters left a comment

Choose a reason for hiding this comment

cgwalters May 4, 2023

Choose a reason for hiding this comment

dbnicholson May 8, 2023

Choose a reason for hiding this comment

cgwalters May 5, 2023

Choose a reason for hiding this comment

dbnicholson May 8, 2023

Choose a reason for hiding this comment

cgwalters May 8, 2023

Choose a reason for hiding this comment

cgwalters May 5, 2023

Choose a reason for hiding this comment

dbnicholson May 8, 2023

Choose a reason for hiding this comment

cgwalters May 8, 2023 • edited Loading

Choose a reason for hiding this comment

cgwalters left a comment

Choose a reason for hiding this comment

cgwalters commented May 5, 2023

cgwalters commented May 5, 2023

cgwalters May 8, 2023 •

edited

Loading