(Urgent!) [release][docs][analyzer] Add 11.0.0. release notes

[Szelethus]

Summary

Late as always, but it seems like not too late :) I tried to add everyone who contributed to the static analyzer in a meaningful way, but if you did and don't find yourself on the reviewer list, please add yourself! I compiled these notes from the following list:

git log llvmorg-11-init..llvmorg-11.0.0-rc2 --oneline -- clang/lib/StaticAnalyzer/ clang/include/clang/StaticAnalyzer/

Seems like this is a release to be proud of!

[Szelethus]

@vsavchenko I just realized a significant portion of your work for this release is the following list:

git log llvmorg-11-init..llvmorg-11.0.0-rc2 --oneline -- clang/utils/analyzer/

Is it a correct guess that while your primary audience are the analyzer developers, it wouldn't hurt to mention it in the release notes?

[martong]

clang/docs/ReleaseNotes.rst

473

Umm, this is still an alpha command line option, plus we improved only the pre-condition checks.
So, I'd rather say something like:

Improve the pre-condition modeling of several functions defined in the POSIX standard. This is in alpha currently. To enable, one must explicitly set the ``ModelPOSIX`` argument of the ``apiModeling.StdCLibraryFunctions`` checker.

[whisperity]

Some grammatical fixes and suggestions, inline. I might have absolutely butchered 80-col in the suggestions (thanks Phab for not showing any indication of line length...), so make sure you manually reformat the document before going forward!

clang/docs/ReleaseNotes.rst

448–449

452–453

460–461

The phrasing is incredibly convoluted here, let's put the positive part of the rule that is enforced first.

466

This is a core change, right? How is this list ordered? Perhaps there should be a sort by category, first the core changes, then the individual checkers' improvements.

473–474

Maybe it's worth mentioning the checker here too, akin to the other list elements?

482

Typo.

482–483

483

Typo.

489–490

492–493

What's the proper way of naming this feature, @martong @dkrupp @xazax.hun? Is it like this, or should it be in capitals, or should it be cross-translation unit?

[whisperity]

clang/docs/ReleaseNotes.rst

439–441

... here.

512

@Szelethus Speaking of labels in the dependency patch D86532, there is no label for the CSA changeset...

[martong]

clang/docs/ReleaseNotes.rst

492–493

I suppose cross-translation is the grammatically correct (?). We tend to refer to the feature, however, with all capitals: Cross Translation Unit. This way the CTU abbreviation is immediate.
Perhaps we should write then Cross Translation Unit (CTU) analysis to be the most precise.

[vsavchenko]

In D86533#2236009, @Szelethus wrote:

@vsavchenko I just realized a significant portion of your work for this release is the following list:

git log llvmorg-11-init..llvmorg-11.0.0-rc2 --oneline -- clang/utils/analyzer/

Is it a correct guess that while your primary audience are the analyzer developers, it wouldn't hurt to mention it in the release notes?

Sure, I think that we can mention that there is a beta-version of the dockerized testing system.

On the other note, a huge portion of my work is related to the solver, like b13d9878b8dc and e63b488f2755. I think it can be mentioned in the notes.
Also, 239c53b72b18 might be pretty important for our Obj-C users.

[Szelethus]

Fixes according to reviewer comments!

clang/docs/ReleaseNotes.rst

473

I don't believe in advertising development only options much, especially that we don't expect, nor want non-developers to interact with apiModeling stuff often. :/ I ended up removing the entry as such, if you don't mind.

473–474

We don't want users to tinker with developer-only checkers.

512

It is only important if you want to use that label -- in fact, this is the only one in the entire file :^)

[martong]

LGTM! (At least those changes that involved me, perhaps it is better to wait for more LGs.)

clang/docs/ReleaseNotes.rst

473

OK, fair enough.

[whisperity]

clang/docs/ReleaseNotes.rst

509

Let's put this into monospace font too. Otherwise LGTM from a readability standpoint.

[steakhal]

clang/docs/ReleaseNotes.rst

487

Unfortunately, my taint related patched not upstreamed yet. So you should not mention that IMO.

[Szelethus]

I'd prefer to have a couple more green lights. In particular, another look on the constraint manager and Objective C stuff would be great.

clang/docs/ReleaseNotes.rst

487

I definitely saw some patches that did land.

[vsavchenko]

In D86533#2248856, @Szelethus wrote:

I'd prefer to have a couple more green lights. In particular, another look on the constraint manager and Objective C stuff would be great.

Those parts are good IMO.

[ASDenysPetrov]

clang/docs/ReleaseNotes.rst

453

I've added the patch "Reasoning about comparison expressions in RangeConstraintManager". You can mention that as well.

[Szelethus]

Added a line about D78933.

clang/docs/ReleaseNotes.rst

453

I added the line you mentioned -- do you like how the notes look now?

[hans]

This has probably had enough time for review. Please commit.

[Szelethus]

Sorry for the slack (which is kind of ironic with my attention grabbing title :) ). Landed in rG791b7e9f73e0064153a7c3db8045a7333a8c390c. Thanks everyone!

[hans]

No worries :) Thanks for the notes!

[ASDenysPetrov]

clang/docs/ReleaseNotes.rst

453

Yes, it's quite good enough.

P.S. Sorry for the late response.

clang/docs/ReleaseNotes.rst

.. code-block:: c++
       do {
         foo();
       } while(1);
 
 libclang
 --------
 
 - ...
 
+.. _release-notes-clang-static-analyzer:
+
 Static Analyzer
 ---------------
 
-- ...
+- Improved the analyzer's understanding of inherited C++ constructors.
+
+- Improved the analyzer's understanding of dynamic class method dispatching in
+  Objective-C.
+
+- Greatly improved the analyzer's constraint solver by better understanding
+  when constraints are imposed on multiple symbolic values that are known to be
+  equal or known to be non-equal. It will now also efficiently reject impossible
+  if-branches between known comparison expressions.
+
+- Added :ref:`on-demand parsing <ctu-on-demand>` capability to Cross Translation
+  Unit (CTU) analysis.
+
+- Numerous fixes and improvements for the HTML output.
+
+- New checker: :ref:`alpha.core.C11Lock <alpha-core-C11Lock>` and
+  :ref:`alpha.fuchsia.Lock <alpha-fuchsia-lock>` checks for their respective
+  locking APIs.
+
+- New checker: :ref:`alpha.security.cert.pos.34c <alpha-security-cert-pos-34c>`
+  finds calls to ``putenv`` where a pointer to an autmoatic variable is passed
+  as an argument.
+
+- New checker: :ref:`webkit.NoUncountedMemberChecker
+  <webkit-NoUncountedMemberChecker>` to enforce the existence of virtual
+  destructors for all uncounted types used as base classes.
+
+- New checker: :ref:`webkit.RefCntblBaseVirtualDtor
+  <webkit-RefCntblBaseVirtualDtor>` checks that only ref-counted types
+  are used as class members, not raw pointers and references to uncounted
+  types.
+
+- New checker: :ref:`alpha.cplusplus.SmartPtr <alpha-cplusplus-SmartPtr>` check
+  for dereference of null smart pointers.
+
+- Moved ``PlacementNewChecker`` out of alpha, making it enabled by default.
+
+- Added support for multi-dimensional variadic arrays in ``core.VLASize``.
+
+- Added a check for insufficient storage in array placement new calls, as well
+  as support for alignment variants to ``cplusplus.PlacementNew``.
+
+- While still in alpha, ``alpha.unix.PthreadLock``, the iterator and container
+  modeling infrastructure, ``alpha.unix.Stream``, and taint analysis were
+  improved greatly. An ongoing, currently off-by-default improvement was made on
+  the pre-condition modeling of several functions defined in the POSIX standard.
+
+- Improved the warning messages of several checkers.
+
+- Fixed a few remaining cases of checkers emitting reports under incorrect
+  checker names, and employed a few restrictions to more easily identify and
+  avoid such errors.
+
+- Moved several non-reporting checkers (those that model, among other things,
+  standard functions, but emit no diagnostics) to be listed under
+  ``-analyzer-checker-help-developer`` instead of ``-analyzer-checker-help``.
+  Manually enabling or disabling checkers found on this list is not supported
+  in production.
+
+- Numerous fixes for crashes, false positives and false negatives in
+  ``unix.Malloc``, ``osx.cocoa.NSError``, and several other checkers.
+
+- Implemented a dockerized testing system to more easily determine the
+  correctness and performance impact of a change to the static analyzer itself.
+  The currently beta-version tool is found in
+  ``(llvm-project repository)/clang/utils/analyzer/SATest.py``.
 
 .. _release-notes-ubsan:
 
 Undefined Behavior Sanitizer (UBSan)
 ------------------------------------
 
 Core Analysis Improvements
 ==========================
 
 - ...