[Webkit-unassigned] [Bug 214142] [WTF] Fix PackedAlignedPtr for X86_64 canonical addresses

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=214142

--- Comment #7 from Mark Lam <mark.lam at apple.com> ---
(In reply to Jim Mason from comment #6)
> (In reply to Mark Lam from comment #5)
> > According to https://www.oracle.com/solaris/technologies/memory.html,
> > addresses with bit 47 set are all reserved for Kernel space.  In fact, the
> > upper bound in user space memory is 0x00008000.00000000 i.e. you should
> > never see an address with bit 47 set in WebKit.  Am I missing something?
> 
> I think you may be looking at SPARC.  X86_64 has the user stack and shared
> objects up there.  It's about half way down:
> 
>       64-bit Kernel's Virtual Memory Layout. (assuming 64 bit app)
> 
>  0xFFFFFD80.00000000  |-----------------------|- KERNELBASE (lower if > 1TB)
>                       |     User stack        |- User space memory
>                       |                       |
>                       | shared objects, etc   |       (grows downwards)
>                       :                       :
>                       |                       |
>  0xFFFF8000.00000000  |-----------------------|
>                       |                       |
>                       | VA Hole / unused      |
>                       |                       |
>  0x00008000.00000000  |-----------------------|
>                       |                       |
>                       |                       |
>                       :                       :
>                       |       user heap       |       (grows upwards)
>                       |                       |
>                       |       user data       |
>                       |-----------------------|
>                       |       user text       |
>  0x00000000.04000000  |-----------------------|
>                       |       invalid         |
>  0x00000000.00000000  +-----------------------+
> 
> 
> I can confirm on Solaris that I see addresses with bit 47 both set and not
> set in PackedAlignedPtr, and if I don't sign-extend the ones with bit 47
> set, it will SIGSEGV.

This was exactly what I was looking at.  Oh, I missed the "User stack and shared objects, etc" section.

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.webkit.org/pipermail/webkit-unassigned/attachments/20200709/8cdeab8f/attachment.htm>