From mboxrd@z Thu Jan 1 00:00:00 1970 From: Ian Kent Subject: Re: Autofs and mount namespaces Date: Sun, 10 Apr 2016 10:53:22 +0800 Message-ID: <1460256802.3508.39.camel@themaw.net> References: <20160407181911.GA9018@vader> <1460076663.3135.37.camel@themaw.net> <20160408202123.GA12948@vader.dhcp.thefacebook.com> Mime-Version: 1.0 Content-Transfer-Encoding: 7bit Return-path: DKIM-Signature: v=1; a=rsa-sha1; c=relaxed/relaxed; d=themaw.net; h=cc :content-transfer-encoding:content-type:date:from:in-reply-to :message-id:mime-version:references:subject:to:x-sasl-enc :x-sasl-enc; s=mesmtp; bh=w1BygN/LOZNE3BeA73M7XhSJVQM=; b=ytEgIf csXX07YB/JZjwrYQ4WjJc2yjsRZthqcop6gWg3WU8hngd1g8O+5yNUhRrClq7zom KDubWxALGPFDVreKCD08kkZH8HlGjhDAwqVllXzrouQJg/GeI/cD8kCDx98ldifa U7vDN8p05OaEP7NORbfgo770El/wWpm9EO79g= DKIM-Signature: v=1; a=rsa-sha1; c=relaxed/relaxed; d= messagingengine.com; h=cc:content-transfer-encoding:content-type :date:from:in-reply-to:message-id:mime-version:references :subject:to:x-sasl-enc:x-sasl-enc; s=smtpout; bh=w1BygN/LOZNE3Be A73M7XhSJVQM=; b=IU1Pdkn2cxBowJ2yAQSPV7cKN/MHNVec2KMZ9y9QDNSOHZ6 GBd8dcpStfB+5aj0M4GOxPx6vZ5VOA3sLhk+nb02vbsOrbiSQhztJNQgIYzvHmUN 9gktMlqwZRpeTAwU2joSXeaFvdcFQW7iJYN5BKNVCaGJd+X+nXJh1bw4XyR8= In-Reply-To: <20160408202123.GA12948@vader.dhcp.thefacebook.com> Sender: autofs-owner@vger.kernel.org List-ID: Content-Type: text/plain; charset="us-ascii" To: Omar Sandoval Cc: autofs@vger.kernel.org, linux-fsdevel@vger.kernel.org, kernel-team@fb.com On Fri, 2016-04-08 at 13:21 -0700, Omar Sandoval wrote: > On Fri, Apr 08, 2016 at 08:51:03AM +0800, Ian Kent wrote: > > On Thu, 2016-04-07 at 11:19 -0700, Omar Sandoval wrote: > > > Hi, Ian, > > > > > > I wanted to bring up the issue of autofs and mount namespaces > > > again, > > > which recently resurfaced in the thread here [1]. In that thread, > > > you > > > mentioned that you had some patches to make autofs namespace-aware > > > that > > > you were holding on to. Do you think you could put those up so we > > > can > > > all work out the issues you mentioned? > > > > I can but I haven't tested them at all. > > > > Possibly I could post them to the autofs list if you want to test > > and > > probably make changes to get them working. > > > > Would that be ok .... umm, I think I need to post the patch now > > anyway > > as it's the easiest way to demonstrate what I think is a better > > approach > > than the patch below .... > > > > I'm struggling to get back to work on this but I'm getting there and > > hope to return to it soon. > > Thanks for taking a look. Your patch fixes the case I provided, and it > does look much better than my hack :) Like you said in your other > email, > there are more cases, but this fix is a good start independent of > those > cases. Were there any other patches that needed testing? It changed a few times since I first looked at it and there were other approaches that had more patches but this single patch is what I have come up with most recently. One problem is that with any change that I can come up with the limit on request call backs (the ELOOP return that's seen) is removed and that means it's possible to get a tight call back loop with no exit condition. For a long time I was most concerned about this but now I'm thinking this is a sufficient improvement, at least for the most common use case and I think a couple of not so common cases, to go ahead with it anyway as a similar call back problem is possible when not using namespaces any way. The daemon does handle that sufficiently well now so ..... Ian -- To unsubscribe from this list: send the line "unsubscribe autofs" in From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from out5-smtp.messagingengine.com ([66.111.4.29]:40296 "EHLO out5-smtp.messagingengine.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750811AbcDJCx2 (ORCPT ); Sat, 9 Apr 2016 22:53:28 -0400 Received: from compute5.internal (compute5.nyi.internal [10.202.2.45]) by mailout.nyi.internal (Postfix) with ESMTP id 392FD20BAD for ; Sat, 9 Apr 2016 22:53:27 -0400 (EDT) Message-ID: <1460256802.3508.39.camel@themaw.net> Subject: Re: Autofs and mount namespaces From: Ian Kent To: Omar Sandoval Cc: autofs@vger.kernel.org, linux-fsdevel@vger.kernel.org, kernel-team@fb.com Date: Sun, 10 Apr 2016 10:53:22 +0800 In-Reply-To: <20160408202123.GA12948@vader.dhcp.thefacebook.com> References: <20160407181911.GA9018@vader> <1460076663.3135.37.camel@themaw.net> <20160408202123.GA12948@vader.dhcp.thefacebook.com> Content-Type: text/plain; charset="UTF-8" Mime-Version: 1.0 Content-Transfer-Encoding: 7bit Sender: linux-fsdevel-owner@vger.kernel.org List-ID: On Fri, 2016-04-08 at 13:21 -0700, Omar Sandoval wrote: > On Fri, Apr 08, 2016 at 08:51:03AM +0800, Ian Kent wrote: > > On Thu, 2016-04-07 at 11:19 -0700, Omar Sandoval wrote: > > > Hi, Ian, > > > > > > I wanted to bring up the issue of autofs and mount namespaces > > > again, > > > which recently resurfaced in the thread here [1]. In that thread, > > > you > > > mentioned that you had some patches to make autofs namespace-aware > > > that > > > you were holding on to. Do you think you could put those up so we > > > can > > > all work out the issues you mentioned? > > > > I can but I haven't tested them at all. > > > > Possibly I could post them to the autofs list if you want to test > > and > > probably make changes to get them working. > > > > Would that be ok .... umm, I think I need to post the patch now > > anyway > > as it's the easiest way to demonstrate what I think is a better > > approach > > than the patch below .... > > > > I'm struggling to get back to work on this but I'm getting there and > > hope to return to it soon. > > Thanks for taking a look. Your patch fixes the case I provided, and it > does look much better than my hack :) Like you said in your other > email, > there are more cases, but this fix is a good start independent of > those > cases. Were there any other patches that needed testing? It changed a few times since I first looked at it and there were other approaches that had more patches but this single patch is what I have come up with most recently. One problem is that with any change that I can come up with the limit on request call backs (the ELOOP return that's seen) is removed and that means it's possible to get a tight call back loop with no exit condition. For a long time I was most concerned about this but now I'm thinking this is a sufficient improvement, at least for the most common use case and I think a couple of not so common cases, to go ahead with it anyway as a similar call back problem is possible when not using namespaces any way. The daemon does handle that sufficiently well now so ..... Ian