Re: retry shm attach for windows (WAS: Re: OK, so culicidae is *still* broken)

From: Amit Kapila <amit(dot)kapila16(at)gmail(dot)com>
To: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
Cc: Petr Jelinek <petr(dot)jelinek(at)2ndquadrant(dot)com>, Robert Haas <robertmhaas(at)gmail(dot)com>, Magnus Hagander <magnus(at)hagander(dot)net>, Michael Paquier <michael(dot)paquier(at)gmail(dot)com>, "Tsunakawa, Takayuki" <tsunakawa(dot)takay(at)jp(dot)fujitsu(dot)com>, Noah Misch <noah(at)leadboat(dot)com>, Andres Freund <andres(at)anarazel(dot)de>, pgsql-hackers <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: retry shm attach for windows (WAS: Re: OK, so culicidae is *still* broken)
Date: 2017-06-05 11:16:33
Message-ID: CAA4eK1K1-JP51ws0h3o45SeZ4T67-OmQBNVHpM4zch-XP6h17w@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

On Mon, Jun 5, 2017 at 9:15 AM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> Amit Kapila <amit(dot)kapila16(at)gmail(dot)com> writes:
>
>> I think the same problem can happen during reattach as well.
>> Basically, MapViewOfFileEx can fail to load image at predefined
>> address (UsedShmemSegAddr).
>
> Once we've successfully done the VirtualAllocEx call, that should hold
> until the VirtualFree call in PGSharedMemoryReAttach, allowing the
> immediately-following MapViewOfFileEx call to succeed. Were that not the
> case, we'd have had problems even without ASLR. We did have problems
> exactly like that before the pgwin32_ReserveSharedMemoryRegion code was
> added.
>

I could not find anything directly in specs which could prove the
theory either way. However, in one of the StackOverflow discussions,
it has been indicated that MapViewOfFile can opt to load the image at
an address different than the predefined address due to ASLR.

> So my feeling about this is that retrying the process creation as
> in my prototype patch ought to be sufficient; if you think it isn't, the
> burden of proof is on you.
>

Sure. I think it is slightly tricky because specs don't say clearly
how ASLR can impact the behavior of any API and in my last attempt I
could not reproduce the issue.

I can try to do basic verification with the patch you have proposed,
but I fear, to do the actual tests as suggested by you is difficult as
it is not reproducible on my machine, but I can still try.

[1] - https://stackoverflow.com/questions/9718616/what-does-mapviewoffile-return/11233456
Refer below text:

"Yes, MapViewOfFile returns the virtual memory base address where the
image has been loaded. The value (content) of this address depends on
whether the image has been successfully loaded at its predefined
address (which has been setup by the linker) or whether the image has
been relocated (because the desired, predefined address is already
occupied or because the image has opt-in to support ASLR)."

--
With Regards,
Amit Kapila.
EnterpriseDB: http://www.enterprisedb.com

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Magnus Hagander 2017-06-05 11:28:38 Re: retry shm attach for windows (WAS: Re: OK, so culicidae is *still* broken)
Previous Message Etsuro Fujita 2017-06-05 09:25:26 Re: Bug in ExecModifyTable function and trigger issues for foreign tables