Re: memory mapped files risks(?)
Re: memory mapped files risks(?)
- Subject: Re: memory mapped files risks(?)
- From: "Quinn \"The Eskimo!\"" <email@hidden>
- Date: Tue, 3 Jan 2012 12:08:37 +0000
On 2 Jan 2012, at 04:07, José Cornado wrote:
> Are there any problems with using memory-mapped files to cross boundary (kernel<->user)?
There's nothing fundamentally broken about sharing a memory mapped file between the kernel and user space. However, as with all shared mappings, you have to be very careful because:
o doing this correctly is difficult, primarily because of the potential race conditions
o there are many potential security issues
o your kernel code must guarantee to not trigger a page fault on the page fault path
This last point is key: keep in mind that all disk and network I/O might potentially be on the page fault path.
* * *
Perhaps you'd like to post some details of your high-level goal. You've asked a bunch of questions that are very much on the edge of what we support for third party KEXTs. If you can give us more context, we should be able to offer better suggestions.
Share and Enjoy
--
Quinn "The Eskimo!" <http://www.apple.com/developer/>
Apple Developer Relations, Developer Technical Support, Core OS/Hardware
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Darwin-kernel mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden