Re: [HACKERS] [PATCHES] fork/exec patch - Mailing list pgsql-hackers-win32

From Merlin Moncure
Subject Re: [HACKERS] [PATCHES] fork/exec patch
Date
Msg-id 303E00EBDD07B943924382E153890E5434AA31@cuthbert.rcsinc.local
Whole thread Raw
List pgsql-hackers-win32
Steve Tibbett wrote:
> code involved - it's a bit of a hack (well, a lot of a hack).  The
> driver
> is writing a value at a fixed undocumented offset from the start
> of the kernel mode thread structure to force it to be alertable.
>
> Seems likely to break at any time if it isn't already broken by
> Windows Server 2003.. also it would mean you'd have to re-create the
> driver for the 64 bit architectures.
>
> So I wouldn't recommend it.. sorry for not checking into it further
> before posting about it.

Yep, I looked at the code too and came to the same conclusion.  Of
course, if Windows was open source, the driver could be compiled for the
platform and there would be no issues.  Also, since most people don't
have the Microsoft ddk the binary driver would have to be distributed,
which is another negative.

Oh well.  Still, if there was no other solution I would still consider
it.

Merlin

pgsql-hackers-win32 by date:

Previous
From: "Steve Tibbett"
Date:
Subject: Re: [HACKERS] [PATCHES] fork/exec patch
Next
From: "Steve Tibbett"
Date:
Subject: Thread-safe signals