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] [thread-next>] [day] [month] [year] [list]
Date:	Tue, 3 Nov 2015 10:01:49 +0800
From:	Wei Yang <weiyang@...ux.vnet.ibm.com>
To:	Alexander Duyck <alexander.duyck@...il.com>
Cc:	Wei Yang <weiyang@...ux.vnet.ibm.com>,
	ethan zhao <ethan.zhao@...cle.com>,
	Bjorn Helgaas <bhelgaas@...gle.com>,
	Alexander Duyck <aduyck@...antis.com>,
	linux-pci@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH v2 5/7] PCI: Wait 1 second between disabling VFs and
 clearing NumVFs

On Mon, Nov 02, 2015 at 07:46:24AM -0800, Alexander Duyck wrote:
>On 11/02/2015 12:33 AM, Wei Yang wrote:
>>On Fri, Oct 30, 2015 at 08:57:17AM -0700, Alexander Duyck wrote:
>>>On 10/29/2015 11:00 PM, ethan zhao wrote:
>>>>Wei,
>>>>
>>>>On 2015/10/30 13:14, Wei Yang wrote:
>>>>>On Thu, Oct 29, 2015 at 05:23:22PM -0500, Bjorn Helgaas wrote:
>>>>>>From: Alexander Duyck <aduyck@...antis.com>
>>>>>>
>>>>>>Per sec 3.3.3.1 of the SR-IOV spec, r1.1, we must allow 1.0s after
>>>>>>clearing
>>>>>>VF Enable before reading any field in the SR-IOV Extended Capability.
>>>>>>
>>>>>>Wait 1 second before calling pci_iov_set_numvfs(), which reads
>>>>>>PCI_SRIOV_VF_OFFSET and PCI_SRIOV_VF_STRIDE after it sets
>>>>>>PCI_SRIOV_NUM_VF.
>>>>>>
>>>>>>[bhelgaas: split to separate patch for reviewability, add spec
>>>>>>reference]
>>>>>>Signed-off-by: Alexander Duyck <aduyck@...antis.com>
>>>>>>Signed-off-by: Bjorn Helgaas <bhelgaas@...gle.com>
>>>>>>---
>>>>>>drivers/pci/iov.c |    2 +-
>>>>>>1 file changed, 1 insertion(+), 1 deletion(-)
>>>>>>
>>>>>>diff --git a/drivers/pci/iov.c b/drivers/pci/iov.c
>>>>>>index fada98d..24428d5 100644
>>>>>>--- a/drivers/pci/iov.c
>>>>>>+++ b/drivers/pci/iov.c
>>>>>>@@ -339,13 +339,13 @@ failed:
>>>>>>    iov->ctrl &= ~(PCI_SRIOV_CTRL_VFE | PCI_SRIOV_CTRL_MSE);
>>>>>>    pci_cfg_access_lock(dev);
>>>>>>    pci_write_config_word(dev, iov->pos + PCI_SRIOV_CTRL, iov->ctrl);
>>>>>>-    pci_iov_set_numvfs(dev, 0);
>>>>>>    ssleep(1);
>>>>>>    pci_cfg_access_unlock(dev);
>>>>>>
>>>>>>    if (iov->link != dev->devfn)
>>>>>>        sysfs_remove_link(&dev->dev.kobj, "dep_link");
>>>>>>
>>>>>>+    pci_iov_set_numvfs(dev, 0);
>>>>>One small question, any specific reason put it here instead of just after
>>>>>sleep()?
>>>>Agree,  pci_iov_set_numvfs(dev, 0) should be put before
>>>>pci_cfg_access_unlock(dev) to avoid race,  because "NumVFs may only be
>>>>written while VF Enable is Clear"
>>>We are already guaranteeing that aren't we?  I'm assuming there is already
>>>code in place here somewhere that prevents us from both enabling and
>>>disabling SR-IOV from more than one thread.  Otherwise how could we hope to
>>>have any sort of consistent state?
>>>
>>>I'm fine with us being more explicit about it if we want to be, but if we are
>>>going to do it we should probably update all 3 spots where we update NumVFs
>>>after init instead of just this one.  Perhaps it should be a separate patch.
>>>
>>Yep, I think the statement is met, "NumVFs may only be written while VF Enable
>>is Clear".
>>
>>While in your commit log, the purpose of this patch is to wait 1 second before
>>write NumVFs. So I am interesting to know why you move this out of the
>>pci_cfg_access_lock. Because it looks better? have better performance?
>>
>>Actually, this is a question instead of a challenge :-)
>
>It is because the first call to pci_iov_set_numvfs is done outside of the
>pci_cfg_access_lock.  This way when I add the clean-up for the bus numbering
>failure in patch 7 I don't have to modify as much code either since the write
>is already pulled out.
>
>An added bonus is the code is now much closer to what we have in
>sriov_disable which has seen much more use than the exception handling case
>for sriov_enable, so it has been more thoroughly tested.

Thanks~ I am more comfortable with this change~

>
>- Alex

-- 
Richard Yang
Help you, Help me

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

Powered by blists - more mailing lists