public inbox for libabigail@sourceware.org
 help / color / mirror / Atom feed
* [Bug default/29441] New: name demangling should not be done on variable names
@ 2022-08-03 12:28 gprocida at google dot com
  2022-08-03 12:39 ` [Bug default/29441] " gprocida at google dot com
  2022-08-03 14:32 ` gprocida at google dot com
  0 siblings, 2 replies; 3+ messages in thread
From: gprocida at google dot com @ 2022-08-03 12:28 UTC (permalink / raw)
  To: libabigail

https://sourceware.org/bugzilla/show_bug.cgi?id=29441

            Bug ID: 29441
           Summary: name demangling should not be done on variable names
           Product: libabigail
           Version: unspecified
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: default
          Assignee: dodji at redhat dot com
          Reporter: gprocida at google dot com
                CC: libabigail at sourceware dot org
  Target Milestone: ---

Hi Dodji.

We noticed some odd output with abidw --annotate. Simple C++ code like:

struct A { int x; };
A v;

results in the annotation for A:

<!-- void -->

This is because "v" as a type demangles to "void". There is no way to turn off
this behaviour at the abi::__cxa_demangle level.

The fix is to ensure that *only* mangled names are passed to the function.

All callers to demangle_cplus_mangled_name are at risk of giving incorrect
results, including compare_symbol_name which itself has a few callers.

One issue is that symbol name is copied into linkage_name (if absent) so the
presence of linkage_name cannot be used indicator of a name being mangled.

If linkage_name cannot be relied on, then it may be possible to only attempt to
demangle C++ function linkage names and nothing else.

Giuliano.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

^ permalink raw reply	[flat|nested] 3+ messages in thread

* [Bug default/29441] name demangling should not be done on variable names
  2022-08-03 12:28 [Bug default/29441] New: name demangling should not be done on variable names gprocida at google dot com
@ 2022-08-03 12:39 ` gprocida at google dot com
  2022-08-03 14:32 ` gprocida at google dot com
  1 sibling, 0 replies; 3+ messages in thread
From: gprocida at google dot com @ 2022-08-03 12:39 UTC (permalink / raw)
  To: libabigail

https://sourceware.org/bugzilla/show_bug.cgi?id=29441

--- Comment #1 from gprocida at google dot com ---
We've also found that this sometimes causes XML to be missing elf-symbol-id
links from var-decl to elf-symbol.

In a test case, changing a variable name from "x" (which demangles to "long
long") to "xx" (which is not a valid mangled name) fixed a missing link.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

^ permalink raw reply	[flat|nested] 3+ messages in thread

* [Bug default/29441] name demangling should not be done on variable names
  2022-08-03 12:28 [Bug default/29441] New: name demangling should not be done on variable names gprocida at google dot com
  2022-08-03 12:39 ` [Bug default/29441] " gprocida at google dot com
@ 2022-08-03 14:32 ` gprocida at google dot com
  1 sibling, 0 replies; 3+ messages in thread
From: gprocida at google dot com @ 2022-08-03 14:32 UTC (permalink / raw)
  To: libabigail

https://sourceware.org/bugzilla/show_bug.cgi?id=29441

--- Comment #2 from gprocida at google dot com ---
Comment 2 is incorrect. We've been able to reproduce the second issue with pure
C and names that have no demangling. I'll open a separate bug.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2022-08-03 14:32 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-08-03 12:28 [Bug default/29441] New: name demangling should not be done on variable names gprocida at google dot com
2022-08-03 12:39 ` [Bug default/29441] " gprocida at google dot com
2022-08-03 14:32 ` gprocida at google dot com

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).