[LLVMbugs] [Bug 22442] New: -Wpadded may be split in two options

bugzilla-daemon at llvm.org bugzilla-daemon at llvm.org
Tue Feb 3 01:14:04 PST 2015


http://llvm.org/bugs/show_bug.cgi?id=22442

            Bug ID: 22442
           Summary: -Wpadded may be split in two options
           Product: clang
           Version: unspecified
          Hardware: PC
                OS: Linux
            Status: NEW
          Severity: enhancement
          Priority: P
         Component: Frontend
          Assignee: unassignedclangbugs at nondot.org
          Reporter: guillaum.bouchard at gmail.com
                CC: llvmbugs at cs.uiuc.edu
    Classification: Unclassified

Hello.

-Wpadded warns about two issues, padding a struct field and padding at the end
(i.e: the whole struct). See the example:

-------------------
struct Bidule
{
    char b;
    int a;
    char c;
};

--->

test.cpp:4:6: warning: padding struct 'Bidule' with 3 bytes to align 'a'
[-Wpadded]
        int a;
            ^
test.cpp:1:8: warning: padding size of 'Bidule' with 3 bytes to alignment
boundary [-Wpadded]
struct Bidule
--------------------------


Usually padding in the middle is a bad thing because the struct can be "easily"
organized differently to reduce the memory footprint. On the other hand,
padding at the end of a structure is common and there is usually no trivial
fix.

As an example, I ran -Wpadded on a large code base, I found 3 padding of field
which I fixed, and more than one thousand warning messages about padding of
struct for which I cannot do something without complex code refactoring.

It may be possible to split -Wpadding in two options, one which warns about
padding fields (which I found really useful 100% of the time) and the other
which warns about padding the whole struct, which IMHO only need to be
activated when needed.

Thank you.

-- 
You are receiving this mail because:
You are on the CC list for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-bugs/attachments/20150203/e41ea84d/attachment.html>


More information about the llvm-bugs mailing list