linux-kernel - Re: [PATCH v5 3/3] PCI: Introduce the disable_acs

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Date:   Tue, 10 Jul 2018 13:26:18 -0600
From:   Logan Gunthorpe <logang@...tatee.com>
To:     Alex Williamson <alex.williamson@...hat.com>
Cc:     linux-kernel@...r.kernel.org, linux-pci@...r.kernel.org,
        linux-doc@...r.kernel.org, Stephen Bates <sbates@...thlin.com>,
        Christoph Hellwig <hch@....de>,
        Bjorn Helgaas <bhelgaas@...gle.com>,
        Jonathan Corbet <corbet@....net>,
        Ingo Molnar <mingo@...nel.org>,
        Thomas Gleixner <tglx@...utronix.de>,
        "Paul E. McKenney" <paulmck@...ux.vnet.ibm.com>,
        Marc Zyngier <marc.zyngier@....com>,
        Kai-Heng Feng <kai.heng.feng@...onical.com>,
        Frederic Weisbecker <frederic@...nel.org>,
        Dan Williams <dan.j.williams@...el.com>,
        Jérôme Glisse <jglisse@...hat.com>,
        Benjamin Herrenschmidt <benh@...nel.crashing.org>,
        Christian König <christian.koenig@....com>,
        Matthew Wilcox <willy@...radead.org>
Subject: Re: [PATCH v5 3/3] PCI: Introduce the disable_acs_redir parameter



On 10/07/18 01:19 PM, Alex Williamson wrote:
> Note that these devices don't have an ACS capability, so they should
> drop out just as any other device without an ACS capability would.
> Should pci_disable_acs_redir() perhaps issue the pci_warn() for all
> such devices, removing this device specific disable function?

Ok, that sounds like a good idea.


> Kind of cumbersome, and as above, maybe the reverse path is optional.
> I wonder if there's a better callback we should use or if we should not
> rely on quirks providing both.

Well, keep in mind enable_acs() and disable_acs_redir() are not inverse
operations. The disable function is only disabling specific ACS bits to
enable redirect -- which are not the same bits being set by the enable
function.

>>  	{ 0 }
>>  };
>>
>>  int pci_dev_specific_enable_acs(struct pci_dev *dev)
>>  {
>> -	const struct pci_dev_enable_acs *i;
>> +	const struct pci_dev_acs_ops *i;
>>  	int ret;
>>
>> -	for (i = pci_dev_enable_acs; i->enable_acs; i++) {
>> +	for (i = pci_dev_acs_ops; i->enable_acs; i++) {
> 
> Perhaps this would walk via ARRAY_SIZE if we decide one or the other
> callback is optional.

> Test i->disable_acs_redir?

Yes, both points make sense if we start saying the operations are optional.


> static inline version for !CONFIG_PCI_QUIRKS?  Thanks,

Oops, yes, I forgot that.

Logan