Add arch-specific provider; remove PunchthroughProvider#806
Open
jaybosamiya-ms wants to merge 1 commit into
Open
Add arch-specific provider; remove PunchthroughProvider#806jaybosamiya-ms wants to merge 1 commit into
jaybosamiya-ms wants to merge 1 commit into
Conversation
jaybosamiya-ms
force-pushed
the
jayb/arch-specific-provider
branch
2 times, most recently
from
e9dffae to
2542f06
Compare
jaybosamiya-ms
force-pushed
the
jayb/arch-specific-provider
branch
from
2542f06 to
8f4da91
Compare
|
🤖 SemverChecks 🤖 Click for details |
CvvT
reviewed
May 8, 2026
| Ok(()) | ||
| } | ||
| ArchSpecificRegister::GsBase => { | ||
| unsafe { litebox_common_linux::wrgsbase(val) }; |
Contributor
There was a problem hiding this comment.
In kernel mode, we call swapgs at the entry and exit of a syscall. At this point, userspace gsbase is stored in MSR (see https://elixir.bootlin.com/linux/v5.19.17/source/arch/x86/kernel/process_64.c#L763).
Contributor
There was a problem hiding this comment.
Note the issue was found by GPT-5.5
Member
There was a problem hiding this comment.
But it depends on when/where this function is called, right? If the caller is after swapgs and it actually knows that it wants to access which register, this function is correct, right?
Member
There was a problem hiding this comment.
Ah, I see the issue. When this kernel platform code executes, if we do wrfsbase or wrgsbase, it actually directly affects the current gsbase used by the kernel. Is my understanding correct?
Contributor
There was a problem hiding this comment.
we need wrmsr(IA32_KERNEL_GS_BASE, val) here to change the "user" GSBASE here due to swapgs.
| ) -> Result<usize, ArchSpecificError> { | ||
| match reg { | ||
| ArchSpecificRegister::FsBase => Ok(unsafe { litebox_common_linux::rdfsbase() }), | ||
| ArchSpecificRegister::GsBase => Ok(unsafe { litebox_common_linux::rdgsbase() }), |
| Ok(v) => Ok(v), | ||
| Err(e) => Err(litebox::platform::PunchthroughError::Failure(e)), | ||
| ArchSpecificRegister::GsBase => { | ||
| unsafe { litebox_common_linux::wrgsbase(val) }; |
| ) -> Result<usize, ArchSpecificError> { | ||
| match reg { | ||
| ArchSpecificRegister::FsBase => Ok(unsafe { litebox_common_linux::rdfsbase() }), | ||
| ArchSpecificRegister::GsBase => Ok(unsafe { litebox_common_linux::rdgsbase() }), |
wdcui
reviewed
May 14, 2026
Member
wdcui
left a comment
wdcui
left a comment
There was a problem hiding this comment.
I left some comments below. Will have some offline discussion with you.
| // Reaching here means that a shim is attempting to easily run on a platform that | ||
| // fully disallows it. There is no reasonable handling here, so we panic. | ||
| // XXX: should this be ENOSYS? | ||
| panic!() |
Member
There was a problem hiding this comment.
I wouldn't expect an error code conversion function to panic?
| Ok(()) | ||
| } | ||
| ArchSpecificRegister::GsBase => { | ||
| unsafe { litebox_common_linux::wrgsbase(val) }; |
Member
There was a problem hiding this comment.
But it depends on when/where this function is called, right? If the caller is after swapgs and it actually knows that it wants to access which register, this function is correct, right?
| litebox::platform::ArchSpecificRegister::GsBase => { | ||
| // GS base is used internally by this platform to hold the host | ||
| // TLS base across the guest/host fs-gs swap, so it is not | ||
| // directly programmable by the guest. |
Member
There was a problem hiding this comment.
Why would you expect the guest to call this platform trait?
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
This PR removes the last vestiges of the
PunchthroughProvider, moving out architecture-specific register management into its own trait. For now, I've only added support for x86-64's fsbase/gsbase, with platforms deciding which ones they'd like to support and which ones they do not want to support. This slightly expands beyond the only-fsbase behavior that existed in the old code, but acts as a way to improve clarity on usability (indeed, handling these generically will be needed if/when we want a proper WinNT shim, so might as well generalize right now).