Date: Fri, 22 Jun 2007 08:19:59 -0700 From: Nate Lawson <nate@root.org> To: Andriy Gapon <avg@icyb.net.ua> Cc: arch@freebsd.org, lars.engels@0x20.net, Ariff Abdullah <ariff@freebsd.org> Subject: Re: patch: acpi usermode suspend API change Message-ID: <467BE89F.9090302@root.org> In-Reply-To: <467B8732.60503@icyb.net.ua> References: <1182140582.00760687.1182127202@10.7.7.3> <1182453789.00762635.1182442802@10.7.7.3> <1182468188.00762718.1182456601@10.7.7.3> <1182482583.00762779.1182469202@10.7.7.3> <467B8732.60503@icyb.net.ua>
next in thread | previous in thread | raw e-mail | index | archive | help
Andriy Gapon wrote: > on 22/06/2007 02:33 Ariff Abdullah said the following: >> Same issue with sound devices as well, but I've managed to solve it by >> implementing own clone manager which require tiny hook into open() and >> close() along with garbage collector and ability to handle various >> abusive behaviour (while : ; ls /dev/xxx ; done, etc). See >> sys/dev/sound/clone.* . It should be generic enough. > > Maybe I'll say something stupid - I haven't looked at any code nor I > have any patches of my own, but wouldn't it be possible to manage cloned > devices similarly to how file descriptors are managed ? That is, don't > leave unused devices behind, re-use lowest available device id, etc. > I think that this would be a right thing. There are still resource exhaustion issues with that simple approach. Since each copy of my device is single-open, I can refuse the 2nd and future processes that try to open the same device. -- Nate
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?467BE89F.9090302>