Changes to add itypes on undefined functions by john-h-kastner · Pull Request #691 · correctcomputation/checkedc-clang

john-h-kastner · 2021-08-25T22:53:37Z

With this change, 3C will insert itypes on functions without definitions. For example:

void foo(int *);

becomes

void foo(int * : itype(_Ptr<int>));

This should be merged into main after #690, #689, and #688.

kyleheadley

This looks good, though I had couple of comments.
Congratulations on getting more checked types in the tests.

Adds checks to ensure that 3C will not attempt to rewrite in unwritable files to insert array bounds. The test file in this PR contains an example using a function body in an unwritable file which fails on the current main. After changes made to support rewriting in functions declarations without definitions (#691), this will be able to happen even when the functions are not defined, and so it will becomes a more prevalent issue.

john-h-kastner · 2021-08-30T18:51:52Z

The new behavior is now behind a flag. This now also depends on #692

This reverts part of commit e8f5ff5.

Fixes an assertion that would fail when attempting to extract the string representation of an itype expression from the original source code if the itype expression was inside a macro. The string representation for itypes in macros is now re-generated from the AST.

mattmccutchen-cci

The flag definition itself looks OK modulo my comment, but there have been so many merges and force-pushes since Kyle's approval that I can't claim to have reviewed the other changes you've made since then. Are we going to just trust that those changes are OK? Or if we want, I can try to construct the right diff to review.

john-h-kastner · 2021-08-31T19:12:28Z

Or if we want, I can try to construct the right diff to review.

The relevant changes should just be d589534...0ec32e0

The force-pushing can safely be ignored.

Co-authored-by: Matt McCutchen (Correct Computation) <matt@correctcomputation.com>

mattmccutchen-cci · 2021-08-31T20:32:24Z

The relevant changes should just be d589534...0ec32e0

Thanks. @kyleheadley I think it makes the most sense if you review these changes since you reviewed the previous version of the same code.

kyleheadley · 2021-09-01T15:19:13Z

The relevant changes should just be d589534...0ec32e0

In this block I see:

ConstraintVariables ~2260:
-  bool SrcChecked = ExternalConstraint->isOriginallyChecked() ||
-                   ExternalConstraint->srcHasItype();
+  bool HasItype = ExternalConstraint->srcHasItype();

But it doesn't appear in the collective change. It's the only code from the block that I have any issue with. Was it perhaps reversed because of how github deals with diffs? Was this a real change and if so what is it for?

john-h-kastner · 2021-09-01T17:34:10Z

Was it perhaps reversed because of how github deals with diffs? Was this a real change and if so what is it for?

This is undoing a change that was present in the original version of the PR

mattmccutchen-cci

One thing I noticed. I'm still hoping Kyle will review all the changes since his previous approval.

Also, please merge main into this branch per our discussion. We expect that to make no difference, but it's good to rule out any surprises during the PR merge.

mattmccutchen-cci · 2021-09-07T14:25:59Z

+  // If we've seen this symbol, but never seen a body for it, constrain
+  // everything about it.


It seems this comment is no longer quite accurate when InferTypesForUndefs is on.

kyleheadley · 2021-09-07T15:26:28Z

One thing I noticed. I'm still hoping Kyle will review all the changes since his previous approval.

Did I not do that in my previous comment?

If there's some confusion about what needs to be reviewed, please send me a link, like was done above, or let me know that I'll need to look over the whole thing again. Otherwise, I'll stick with trusting that our team wants to have their code reviewed and has made their best effort to provide it. Since, as far I as recall, "submitter presses merge" and takes responsibility for their own code.

mattmccutchen-cci · 2021-09-07T15:29:04Z

One thing I noticed. I'm still hoping Kyle will review all the changes since his previous approval.

Did I not do that in my previous comment?

If there's some confusion about what needs to be reviewed, please send me a link, like was done above, or let me know that I'll need to look over the whole thing again. [...]

My concern was that that discussion had not reached a conclusion (as could be expressed by a GitHub re-approval), not that there was confusion about what needed review. Maybe I should have said "approve" rather than "review". 🙂

kyleheadley

re-approve (code so far). Carry on.

kyleheadley approved these changes Aug 26, 2021

View reviewed changes

Comment thread clang/lib/3C/DeclRewriter.cpp Outdated

Comment thread clang/lib/3C/ProgramInfo.cpp Outdated

Comment thread clang/test/3C/function_typedef.c Outdated

john-h-kastner force-pushed the rewrite_extern branch from e7481b9 to e0dbee0 Compare August 27, 2021 14:03

john-h-kastner force-pushed the rewrite_extern.base branch from f0f6fc0 to 4b90959 Compare August 27, 2021 14:04

john-h-kastner force-pushed the rewrite_extern branch from e0dbee0 to 5855166 Compare August 27, 2021 19:26

john-h-kastner changed the base branch from rewrite_extern.base to main August 27, 2021 19:27

john-h-kastner force-pushed the rewrite_extern branch from 5855166 to c434414 Compare August 30, 2021 18:50

john-h-kastner changed the base branch from main to options_cleanup August 30, 2021 18:50

john-h-kastner added 6 commits August 30, 2021 15:35

fix typedef function rewriting errors

9396d76

Changes to add itypes on undefined functions

3025472

Hide undefined function rewriting behind flag

987611f

Revert test changes from "Changes to add itypes on undefined functions"

d589534

This reverts part of commit e8f5ff5.

Reduce duplicate code in ProgramInfo::link

1e3e262

Update expected root cause text

c019fe6

john-h-kastner force-pushed the rewrite_extern branch from c434414 to c019fe6 Compare August 30, 2021 19:54

john-h-kastner changed the base branch from options_cleanup to main August 30, 2021 19:55

john-h-kastner added 4 commits August 30, 2021 16:45

Undo obsolete change that was needed before other implementation changes

a205b0f

Expand test coverage

0ec32e0

Comment edits

cc9024a

mattmccutchen-cci reviewed Aug 31, 2021

View reviewed changes

Comment thread clang/tools/3c/3CStandalone.cpp Outdated

john-h-kastner and others added 2 commits August 31, 2021 15:14

Update clang/tools/3c/3CStandalone.cpp

fde7bda

Co-authored-by: Matt McCutchen (Correct Computation) <matt@correctcomputation.com>

Update test

524eb39

mattmccutchen-cci requested a review from kyleheadley August 31, 2021 20:32

Refernce issue

721358d

john-h-kastner added 3 commits September 2, 2021 10:41

Add test case from #594

46e5a3d

Merge branch 'itype_in_macro' into rewrite_extern

26553d1

Merge remote-tracking branch 'origin/main' into rewrite_extern

e17cd94

mattmccutchen-cci reviewed Sep 7, 2021

View reviewed changes

kyleheadley approved these changes Sep 7, 2021

View reviewed changes

Update comment

f75979d

john-h-kastner merged commit 584c6ba into main Sep 8, 2021

This was referenced Sep 9, 2021

liberal itypes for prototypes (with no def) #402

Closed

Maintaining checked versions of headers for an external library #401

Open

kyleheadley mentioned this pull request Sep 14, 2021

First draft of itypes on prototypes #435

Draft

Machiry deleted the rewrite_extern branch January 22, 2022 15:37

		// If we've seen this symbol, but never seen a body for it, constrain
		// everything about it.

Conversation

john-h-kastner commented Aug 25, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

kyleheadley left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

john-h-kastner commented Aug 30, 2021

Uh oh!

mattmccutchen-cci left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

john-h-kastner commented Aug 31, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

mattmccutchen-cci commented Aug 31, 2021

Uh oh!

kyleheadley commented Sep 1, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

john-h-kastner commented Sep 1, 2021

Uh oh!

mattmccutchen-cci left a comment

Choose a reason for hiding this comment

Uh oh!

mattmccutchen-cci Sep 7, 2021

Choose a reason for hiding this comment

Uh oh!

kyleheadley commented Sep 7, 2021

Uh oh!

mattmccutchen-cci commented Sep 7, 2021

Uh oh!

kyleheadley left a comment

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

john-h-kastner commented Aug 25, 2021 •

edited

Loading

kyleheadley left a comment •

edited

Loading

john-h-kastner commented Aug 31, 2021 •

edited

Loading

kyleheadley commented Sep 1, 2021 •

edited

Loading