[Webkit-unassigned] [Bug 127902] Fix the remaining regression caused by the jsCStack branch merge on Linux platforms

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=127902





--- Comment #16 from Michael Saboff <msaboff at apple.com>  2014-02-25 16:48:06 PST ---
(In reply to comment #15)
> Created an attachment (id=225196)
 --> (https://bugs.webkit.org/attachment.cgi?id=225196&action=review) [details]
> Debugging data - info proc all, instructions, registers with ToT
> 
> This text file contains GDB output of the `info proc all` command. I hope that properly represents the requested memory mappings.
> 
> There's also the call_varargs instruction list and register values I produced with a ToT build.

In the attachment, the memory map shows:
          Start Addr           End Addr       Size     Offset objfile
      0x7ffffff7b000     0x7ffffffff000    0x84000        0x0 [stack]

And the registers show:
Registers:
rdx            0x7fffffefc380    140737487291264
rbp            0x7ffffff7c400    0x7ffffff7c400
rsp            0x7ffffff7c380    0x7ffffff7c380

rdx - 0x7fffffefc380 - Isn't part of the mapped address space.  The current stack allocation isn't big enough for this call.

The value pointed to by VM::m_jsStackLimit is also outside of the allocated stack area.
%r11:
0x4537c0:    0x00007fffffc1df20

Could it be that when VM::m_jsStackLimit was updated, the stack allocation wasn't increased?  Or maybe VM:m_jsStack shouldn't have been updated.

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.