[Lldb-commits] Proposed small tweak to the Profile Until Ret Instruction patch

Todd Fiala tfiala at google.com
Mon Aug 25 16:19:15 PDT 2014


svn commit
Sending        source/Symbol/FuncUnwinders.cpp
Transmitting file data .
Committed revision 216416.

Tested on Linux and MacOSX.  All tests passed.


On Mon, Aug 25, 2014 at 4:11 PM, Tong Shen <endlessroad at google.com> wrote:

> Hi Todd,
>
> My patch alone. Confirmed.
> Possible commit message: "Copy unwind plan instead of modifying it
> directly, so "image show-unwind" prints different plans for asynchronous
> and synchronous"
>
>
> On Mon, Aug 25, 2014 at 3:25 PM, Tong Shen <endlessroad at google.com> wrote:
>
>> Got it :-)
>>
>>
>> On Mon, Aug 25, 2014 at 3:24 PM, Jason Molenda <jmolenda at apple.com>
>> wrote:
>>
>>> Awesome, thanks, I'll commit my patch.  Please commit your patch to copy
>>> the UnwindPlan instead of modifying in place when you're ready.
>>>
>>>
>>> > On Aug 25, 2014, at 3:13 PM, Tong Shen <endlessroad at google.com> wrote:
>>> >
>>> > Hi Jason,
>>> >
>>> > Patch confirmed, plan name changed as expected (and did not change for
>>> gcc because gcc describes epilogue as well).
>>> >
>>> > I made a small change though: now that you brought up "image dump
>>> show-unwind", I found that I should make a copy of
>>> m_unwind_plan_call_site_sp instead of modifying it directly.
>>> > Now "image show-unwind" output looks like:
>>> >
>>> > Asynchronous (not restricted to call-sites) UnwindPlan for 1`f (start
>>> addr 0x8048420):
>>> > This UnwindPlan originally sourced from eh_frame CFI plus augmentation
>>> from assembly parsing
>>> > Address range of this UnwindPlan: [1..text + 256-0x0000014b)
>>> > row[0]: 0x00000000: CFA=esp +4 => eip=[esp]
>>> > row[1]: 0x00000003: CFA=esp+24 => eip=[esp+20]
>>> > row[2]: 0x0000004a: CFA=esp +4 => eip=[esp]
>>> >
>>> > Synchronous (restricted to call-sites) UnwindPlan for 1`f (start addr
>>> 0x8048420):
>>> > This UnwindPlan originally sourced from eh_frame CFI
>>> > Address range of this UnwindPlan: [1..text + 256-0x0000014b)
>>> > row[0]: 0x00000000: CFA=esp +4 => eip=[esp]
>>> > row[1]: 0x00000003: CFA=esp+24 => eip=[esp+20]
>>> >
>>> > Architecture default UnwindPlan for 1`f (start addr 0x8048420):
>>> > This UnwindPlan originally sourced from i386 default unwind plan
>>> > row[0]: 0x00000000: CFA=ebp +8 => esp=ebp+8 ebp=[ebp] eip=[ebp+4]
>>> >
>>> >
>>> >
>>> > On Mon, Aug 25, 2014 at 2:45 PM, Tong Shen <endlessroad at google.com>
>>> wrote:
>>> > Of course :-)
>>> > Keeping the old name may cause confusion in the future.
>>> >
>>> > I will test it now and make sure it works.
>>> >
>>> > Thanks!
>>> >
>>> >
>>> > On Mon, Aug 25, 2014 at 2:43 PM, Jason Molenda <jmolenda at apple.com>
>>> wrote:
>>> > Hi Tong, I think it might be useful to note that the UnwindPlan has
>>> been modified (in case one of us is looking at it with "image dump
>>> show-unwind").  What do you think about this patch?  I don't have eh_frame
>>> info on my system right now so I haven't tested this.
>>> >
>>> >
>>> >
>>> >
>>> > --
>>> > Best Regards, Tong Shen
>>> >
>>> >
>>> >
>>> > --
>>> > Best Regards, Tong Shen
>>> > <unwindplan-name_v2.txt>
>>>
>>>
>>
>>
>> --
>> Best Regards, Tong Shen
>>
>
>
>
> --
> Best Regards, Tong Shen
>



-- 
Todd Fiala | Software Engineer | tfiala at google.com | 650-943-3180
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/lldb-commits/attachments/20140825/12924e52/attachment.html>


More information about the lldb-commits mailing list