Documentation: livepatch: document reliable stacktrace
Subsystems (2)
PATCH v4 1/1 Documentation: livepatch: document reliable stacktrace
From: Mark Rutland <[email protected]>
Add documentation for reliable stacktrace. This is intended to describe the semantics and to be an aid for implementing architecture support for HAVE_RELIABLE_STACKTRACE.
Unwinding is a subtle area, and architectures vary greatly in both implementation and the set of concerns that affect them, so I've tried to avoid making this too specific to any given architecture. I've used examples from both x86_64 and arm64 to explain corner cases in more detail, but I've tried to keep the descriptions sufficient for those who are unfamiliar with the particular architecture.
I've tried to give rationale for all the recommendations/requirements, since that makes it easier to spot nearby issues, or when a check happens to catch a few things at once. I believe what I have written is sound, but as some of this was reverse-engineered I may have missed things worth noting.
I've made a few assumptions about preferred behaviour, notably:
* If you can reliably unwind through exceptions, you should (as x86_64 does).
* It's fine to omit ftrace_return_to_handler and other return trampolines so long as these are not subject to patching and the original return address is reported. Most architectures do this for ftrace_return_handler, but not other return trampolines.
* For cases where link register unreliability could result in duplicate entries in the trace or an inverted trace, I've assumed this should be treated as unreliable. This specific case shouldn't matter to livepatching, but I assume that that we want a reliable trace to have the correct order.
--- Documentation/livepatch/index.rst | 1 + .../livepatch/reliable-stacktrace.rst | 318 ++++++++++++++++++ 2 files changed, 319 insertions(+) create mode 100644 Documentation/livepatch/reliable-stacktrace.rst
v4:
- Renumber table of contents
v3:
- Incorporated objtool section from Mark.
- Deleted confusing notes about using annotations.
Documentation/livepatch/index.rst +1 -0
@@ -13,6 +13,7 @@ Kernel Livepatching module-elf-format shadow-vars system-state + reliable-stacktrace .. only:: subproject and html
Documentation/livepatch/reliable-stacktrace.rst new file +318 -0
318 lines not on this page · raw message
Signature · 1 line
2.20.1
Petr Mladek
maintainerHi Mark,
first, thanks a lot for writing this.
Quoted 13 lines · Mark Brown
From: Mark Rutland <[email protected]>
Add documentation for reliable stacktrace. This is intended to describe the semantics and to be an aid for implementing architecture support for
HAVE_RELIABLE_STACKTRACE.Unwinding is a subtle area, and architectures vary greatly in both implementation and the set of concerns that affect them, so I've tried to avoid making this too specific to any given architecture. I've used examples from both
x86_64and arm64 to explain corner cases in more detail, but I've tried to keep the descriptions sufficient for those who are unfamiliar with the particular architecture.
Quoted 3 lines
I've tried to give rationale for all the recommendations/requirements, since that makes it easier to spot nearby issues, or when a check happens to catch a few things at once.
The above looks enough for the commit message. Well, Josh, typically asks for a directive style, example:
Instead of "I've tried to give rationale...", please use something like "The documentation gives rationale...".
Quoted 14 lines
I believe what I have written is
sound, but as some of this was reverse-engineered I may have missed things worth noting.I've made a few assumptions about preferred behaviour, notably:
* If you can reliably unwind through exceptions, you should (as
x86_64does).* It's fine to omit
ftrace_return_to_handlerand other return trampolines so long as these are not subject to patching and the original return address is reported. Most architectures do this forftrace_return_handler, but not other return trampolines.
Quoted 5 lines
* For cases where link register unreliability could result in duplicate entries in the trace or an inverted trace, I've assumed this should be treated as unreliable. This specific case shouldn't matter to livepatching, but I assume that that we want a reliable trace to have the correct order.
This looks like a background that is typically part of the cover leter. It mentions some Mark's doubts.
Could anyone please answer whether the above assumptions are correct or not? We should remove them from the commit message. If any assumption is wrong, we should fix the documentation.
Honestly, I am curious about the anwer. I am not familiar with these details of the reliable stacktrace ;-)
Otherwise, it looks good to me on the first look. But I am not expert in this are, so I could not check the details effectively.
Best Regards,
Petr
Quoted 22 lines · Petr Mladek
Hi Mark,
first, thanks a lot for writing this.
Quoted 13 lines · Mark Brown
From: Mark Rutland <[email protected]>
Add documentation for reliable stacktrace. This is intended to describe the semantics and to be an aid for implementing architecture support for
HAVE_RELIABLE_STACKTRACE.Unwinding is a subtle area, and architectures vary greatly in both implementation and the set of concerns that affect them, so I've tried to avoid making this too specific to any given architecture. I've used examples from both
x86_64and arm64 to explain corner cases in more detail, but I've tried to keep the descriptions sufficient for those who are unfamiliar with the particular architecture.Quoted 3 lines
I've tried to give rationale for all the recommendations/requirements, since that makes it easier to spot nearby issues, or when a check happens to catch a few things at once.
Quoted 5 lines
The above looks enough for the commit message. Well, Josh, typically asks for a directive style, example:
Instead of "I've tried to give rationale...", please use something like "The documentation gives rationale...".
True, we do try to use imperative form like "Try to give rationale...".
Though documentation is less technical than code, so maybe technical language is less important.
Quoted 20 lines
Quoted 14 lines
I believe what I have written is
sound, but as some of this was reverse-engineered I may have missed things worth noting.I've made a few assumptions about preferred behaviour, notably:
* If you can reliably unwind through exceptions, you should (as
x86_64does).* It's fine to omit
ftrace_return_to_handlerand other return trampolines so long as these are not subject to patching and the original return address is reported. Most architectures do this forftrace_return_handler, but not other return trampolines.Quoted 5 lines
* For cases where link register unreliability could result in duplicate entries in the trace or an inverted trace, I've assumed this should be treated as unreliable. This specific case shouldn't matter to livepatching, but I assume that that we want a reliable trace to have the correct order.
Quoted 6 lines
This looks like a background that is typically part of the cover leter. It mentions some Mark's doubts.
Could anyone please answer whether the above assumptions are correct or not? We should remove them from the commit message. If any assumption is wrong, we should fix the documentation.
Agreed, this section can probably be dropped.
Signature · 1 line
Josh
Hi Petr,
Quoted 5 lines · Petr Mladek
Quoted 4 lines · Mark Brown
I've made a few assumptions about preferred behaviour, notably:
* If you can reliably unwind through exceptions, you should (as
x86_64does).
IIRC this was confirmed as desireable, and the text already reflects this.
Quoted 4 lines
Quoted 4 lines
* It's fine to omit
ftrace_return_to_handlerand other return trampolines so long as these are not subject to patching and the original return address is reported. Most architectures do this forftrace_return_handler, but not other return trampolines.
Likewise I think we agreed this was fine, given these were not themselves subkect to patching.
Quoted 5 lines
Quoted 5 lines
* For cases where link register unreliability could result in duplicate entries in the trace or an inverted trace, I've assumed this should be treated as unreliable. This specific case shouldn't matter to livepatching, but I assume that that we want a reliable trace to have the correct order.
I don't think we had any comments either way on this, but I think it's sane to say this for now and later relax it if we need to.
... so I reckon we can just delete all this as Josh suggests. Any acks for the patch itself tacitly agrees with these points. :)
Thanks,
Mark.