linux-kernel - Re: [PATCH 2/2] uio: Prefer MSI(X) interrupts in PCI drivers

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, 24 Oct 2017 15:11:52 +0200
From:   Stephen Hemminger <stephen@...workplumber.org>
To:     "gregkh@...uxfoundation.org" <gregkh@...uxfoundation.org>
Cc:     "Stahl, Manuel" <manuel.stahl@...-extern.fraunhofer.de>,
        "devel@...verdev.osuosl.org" <devel@...verdev.osuosl.org>,
        "hjk@...utronix.de" <hjk@...utronix.de>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "sojkam1@....cvut.cz" <sojkam1@....cvut.cz>
Subject: Re: [PATCH 2/2] uio: Prefer MSI(X) interrupts in PCI drivers

On Fri, 20 Oct 2017 14:50:44 +0200
"gregkh@...uxfoundation.org" <gregkh@...uxfoundation.org> wrote:

> On Fri, Oct 06, 2017 at 07:57:00AM -0700, Stephen Hemminger wrote:
> > On Fri, 6 Oct 2017 13:50:44 +0000
> > "Stahl, Manuel" <manuel.stahl@...-extern.fraunhofer.de> wrote:
> >   
> > > MSI(X) interrupts are not shared between devices. So when available
> > > those should be preferred over legacy interrupts.
> > > 
> > > Signed-off-by: Manuel Stahl <manuel.stahl@....fraunhofer.de>
> > > ---
> > >  drivers/uio/uio_pci_dmem_genirq.c | 27 ++++++++++++++++++++-------
> > >  drivers/uio/uio_pci_generic.c     | 24 ++++++++++++++++++------
> > >  2 files changed, 38 insertions(+), 13 deletions(-)  
> > 
> > The last time I tried to do MSI-X with pci-generic it got rejected
> > by the maintainer.  
> 
> Hm, yeah, this would break users today that do not have msi-x, right?
> 
> Not good, Manuel, how well did you test this?
> 
> thanks,
> 
> greg k-h

Look at https://patchwork.kernel.org/patch/7303021/

The objection was more that UIO developers did not like that
UIO was (already) being used for DMA without IOMMU, and MSI-x has DMA because
of vector table.