[llvm-commits] [PATCH] Fix for bug in exception table allocation

[Michael Muller]

Done. New patch attached.

(I'll update the patch in the bug later)

Was there a recent style normalization effort in the codebase?  (I try to
emulate the style of the context)


On Thu, Jan 10, 2013 at 3:53 PM, Jim Grosbach <grosbach at apple.com> wrote:

> Missed one.
>
> s/CreateMemoryManager/createMemoryManager/
>
> OK to commit with that change.
>
>
> On Jan 10, 2013, at 11:50 AM, Michael Muller <mmuller at google.com> wrote:
>
> Ping?
>
>
> On Tue, Jan 8, 2013 at 6:03 PM, Michael Muller <mmuller at google.com> wrote:
>
>>
>>
>>
>> On Tue, Jan 8, 2013 at 5:52 PM, Jim Grosbach <grosbach at apple.com> wrote:
>>
>>> Looks good to me, with a couple nits:
>>>
>>> +  virtual RecordingJITMemoryManager* CreateMemoryManager() {
>>>
>>> The '*' should bind to the identifier, not the type. The function name
>>> should start with lower case. There's a few other instances in the patch of
>>> the former. Please double-check.
>>>
>>
>> All fixed.
>>
>>
>>>
>>> Thanks!
>>>
>>> -Jim
>>>
>>>
>>> On Jan 8, 2013, at 2:43 PM, Michael Muller <mmuller at google.com> wrote:
>>>
>>>
>>> New patch attached, incorporates Misha's comments and current trunk.
>>>
>>>
>>> On Fri, Jan 4, 2013 at 7:44 PM, Misha Brukman <brukman at gmail.com> wrote:
>>>
>>>> On Wed, Dec 19, 2012 at 1:55 PM, Michael Muller <mmuller at enduden.com>wrote:
>>>>
>>>>> See http://llvm.org/bugs/show_bug.cgi?id=13678 for more history.
>>>>>
>>>>> See also:
>>>>>   http://lists.cs.uiuc.edu/pipermail/llvmdev/2012-August/052766.html
>>>>>
>>>>> http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20120910/150422.html
>>>>>
>>>>> http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20120917/150801.html
>>>>>   http://lists.cs.uiuc.edu/pipermail/llvmdev/2012-December/056829.html
>>>>>
>>>>> When you run out of space writing to a buffer, the buffer management
>>>>> code
>>>>> simply stops writing at the end of the buffer.  It is the
>>>>> responsibility of
>>>>> the caller to verify that it has stayed in bounds and perform a retry
>>>>> with
>>>>> a larger memory estimate if not.  The function writing code does this,
>>>>> but
>>>>> the exception table code following it does not.  The end result is that
>>>>> exception table pointers can get registered pointing to invalid data,
>>>>> causing
>>>>> seg-faults when an exception is thrown.
>>>>>
>>>>
>>>> Minor drive-by style comments (I wish there were a web UI for entering
>>>> these in context):
>>>>
>>>>    - s/;;/;/
>>>>    - ASSERT_TRUE(actual == expected) --> ASSERT_EQ(expected, actual)
>>>>    - add space around binary operators (-, *, etc.), e.g.:
>>>>
>>>> +      ActualSize = (CurBufferPtr-BufferBegin)*2;
>>>>
>>>>
>>> I'm sure I copied that from somewhere, but looking at the file now I see
>>> nothing resembling such formatting.  Fixed.
>>>
>>>
>>>>
>>>>    - fix arg alignment of "ActualSize" here:
>>>>
>>>> +    while (true) {
>>>> +      BufferBegin = CurBufferPtr = MemMgr->startExceptionTable(F.getFunction(),
>>>> +                                                              ActualSize);
>>>>
>>>>
>>> Done.
>>>
>>>
>>>> Misha
>>>>
>>>
>>>
>>>
>>> --
>>> Michael Muller
>>> Ym9yZWQ/
>>> <ExceptionTableOverflowFix-v5.patch>
>>>
>>>
>>>
>>
>>
>> --
>> Michael Muller
>> Ym9yZWQ/
>>
>
>
>
> --
> Michael Muller
> Ym9yZWQ/
>
>
>


-- 
Michael Muller
Ym9yZWQ/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20130110/dcf1f496/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: ExceptionTableOverflowFix-v7.patch
Type: application/octet-stream
Size: 2964 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20130110/dcf1f496/attachment.obj>