Re: [Xenomai] __get_user/__put_user

[Jan Kiszka <jan.kiszka@web.de>]

On 2013-05-09 21:08, Gilles Chanteperdrix wrote:
>> Right, but the general pattern for this is
>>
>> 	obj = xnregistry_fetch(handle)
>> 	get_nklock()
>> 	if (valid(obj))
>> 		do_something_with(obj)
>> 	put_nklock()
> 
> 
> I can do that, it avoids the issue with __ipipe_uaccess_might_fault, but:
> - it is not the same guarantee, nothing tells you that obj is the one
> associated with the name you passed to xnregistry_fetch, it may have
> been destroyed and created in the mean time

See my other reply: user mistake, it should not matter.

> - I am not sure we do not have some other places where we call
> copy_from/to_user with irqs off.

Might be the case, but I think so far we should have tried to avoid
faulting under the nklock as we weren't considering this safe enough or
at least good for the worst case latency.

Jan


-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 263 bytes
Desc: OpenPGP digital signature
URL: <http://www.xenomai.org/pipermail/xenomai/attachments/20130509/5563b2b2/attachment.pgp>