From: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
To: "Pali Rohár" <pali@kernel.org>
Cc: "Marek Behún" <kabel@kernel.org>,
"Bjorn Helgaas" <bhelgaas@google.com>,
linux-pci@vger.kernel.org
Subject: Re: [PATCH 10/14] PCI: aardvark: Enable MSI-X support
Date: Thu, 28 Oct 2021 12:08:35 +0100 [thread overview]
Message-ID: <20211028110835.GA1846@lpieralisi> (raw)
In-Reply-To: <20211027142307.lrrix5yfvroxl747@pali>
On Wed, Oct 27, 2021 at 04:23:07PM +0200, Pali Rohár wrote:
> On Wednesday 27 October 2021 15:12:46 Lorenzo Pieralisi wrote:
> > On Tue, Oct 12, 2021 at 06:41:41PM +0200, Marek Behún wrote:
> > > From: Pali Rohár <pali@kernel.org>
> > >
> > > According to PCI 3.0 specification, sending both MSI and MSI-X interrupts
> > > is done by DWORD memory write operation to doorbell message address. The
> > > write operation for MSI has zero upper 16 bits and the MSI interrupt number
> > > in the lower 16 bits, while the write operation for MSI-X contains a 32-bit
> > > value from MSI-X table.
> > >
> > > Since the driver only uses interrupt numbers from range 0..31, the upper
> > > 16 bits of the DWORD memory write operation to doorbell message address
> > > are zero even for MSI-X interrupts. Thus we can enable MSI-X interrupts.
> >
> > It is the controller driver that defines the MSI-X data field yes, what
> > I don't get is why we have to add this comment in the commit log.
> >
> > Basically Aardvark can support MSI-X up to 32 MSI-X vectors and you
> > are enabling them with this patch.
> >
> > Is there anything *else* I am missing wrt 16-bit/32-bit data fields
> > that we need to know ?
> >
> > > Testing proves that kernel can correctly receive MSI-X interrupts from
> > > PCIe cards which supports both MSI and MSI-X interrupts.
> >
> > I don't understand what you want to convey with this commit log.
> >
> > To me, the whole comment does not add anything (if I understood it),
> > please let me know what you want to express with it.
> >
> > To me this patch enables MSI-X support because the HW can support them,
> > that's it.
>
> My understanding is that MSI-X by definition uses 32-bit write
> operations to doorbell address and so, HW needs to support catching of
> 32-bit write operations.
>
> Aardvark hw seems to support only 16-bit write operation to doorbell
> address. But our testing proved that hw can catch also lower 16-bits of
> 32-bit write operation to doorbell address.
>
> So if driver enforces that every 32-bit write operation to doorbell
> address would have upper 16-bit zeroed then MSI-X should work.
That's clearer than the current commit log.
> In commit message I originally tried to explain it that after applying
> all previous patches which are fixing MSI and Multi-MSI support (part of
> them is enforcement to use only MSI numbers 0..31), it makes driver
> compatible with also MSI-X interrupts.
>
> If you want to rewrite commit message, let us know, there is no problem.
I think we should.
> > > Signed-off-by: Pali Rohár <pali@kernel.org>
> > > Reviewed-by: Marek Behún <kabel@kernel.org>
By the way, this tag should be removed. Marek signed it off, that
applies to other patches in this series as well.
Lorenzo
> > > Signed-off-by: Marek Behún <kabel@kernel.org>
> > > ---
> > > drivers/pci/controller/pci-aardvark.c | 2 +-
> > > 1 file changed, 1 insertion(+), 1 deletion(-)
> > >
> > > diff --git a/drivers/pci/controller/pci-aardvark.c b/drivers/pci/controller/pci-aardvark.c
> > > index b703b271c6b1..337b61508799 100644
> > > --- a/drivers/pci/controller/pci-aardvark.c
> > > +++ b/drivers/pci/controller/pci-aardvark.c
> > > @@ -1274,7 +1274,7 @@ static int advk_pcie_init_msi_irq_domain(struct advk_pcie *pcie)
> > >
> > > msi_di = &pcie->msi_domain_info;
> > > msi_di->flags = MSI_FLAG_USE_DEF_DOM_OPS | MSI_FLAG_USE_DEF_CHIP_OPS |
> > > - MSI_FLAG_MULTI_PCI_MSI;
> > > + MSI_FLAG_MULTI_PCI_MSI | MSI_FLAG_PCI_MSIX;
> > > msi_di->chip = msi_ic;
> > >
> > > pcie->msi_inner_domain =
> > > --
> > > 2.32.0
> > >
next prev parent reply other threads:[~2021-10-28 11:08 UTC|newest]
Thread overview: 36+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-10-12 16:41 [PATCH 00/14] PCI: aardvark controller fixes BATCH 2 Marek Behún
2021-10-12 16:41 ` [PATCH 01/14] PCI: pci-bridge-emul: Fix emulation of W1C bits Marek Behún
2021-10-12 16:41 ` [PATCH 02/14] PCI: aardvark: Fix return value of MSI domain .alloc() method Marek Behún
2021-10-27 11:26 ` Lorenzo Pieralisi
2021-10-27 11:31 ` Pali Rohár
2021-10-12 16:41 ` [PATCH 03/14] PCI: aardvark: Read all 16-bits from PCIE_MSI_PAYLOAD_REG Marek Behún
2021-10-12 16:41 ` [PATCH 04/14] PCI: aardvark: Fix support for MSI interrupts Marek Behún
2021-10-12 16:41 ` [PATCH 05/14] PCI: aardvark: Fix reading MSI interrupt number Marek Behún
2021-10-12 16:41 ` [PATCH 06/14] PCI: aardvark: Clear all MSIs at setup Marek Behún
2021-10-12 16:41 ` [PATCH 07/14] PCI: aardvark: Refactor unmasking summary MSI interrupt Marek Behún
2021-10-12 16:41 ` [PATCH 08/14] PCI: aardvark: Fix masking MSI interrupts Marek Behún
2021-10-12 16:41 ` [PATCH 09/14] PCI: aardvark: Fix setting MSI address Marek Behún
2021-10-12 16:41 ` [PATCH 10/14] PCI: aardvark: Enable MSI-X support Marek Behún
2021-10-27 14:12 ` Lorenzo Pieralisi
2021-10-27 14:23 ` Pali Rohár
2021-10-28 11:08 ` Lorenzo Pieralisi [this message]
2021-10-28 11:13 ` Pali Rohár
2021-10-28 11:30 ` Lorenzo Pieralisi
2021-10-28 11:37 ` Pali Rohár
2021-10-28 15:24 ` Marc Zyngier
2021-10-28 15:29 ` Pali Rohár
2021-10-28 15:51 ` Marek Behún
2021-10-28 16:22 ` Marc Zyngier
2021-10-28 16:25 ` Marek Behún
2021-10-28 17:00 ` Marc Zyngier
2021-10-28 17:47 ` Lorenzo Pieralisi
2021-10-28 18:24 ` Marek Behún
2021-10-12 16:41 ` [PATCH 11/14] PCI: aardvark: Fix support for bus mastering and PCI_COMMAND on emulated bridge Marek Behún
2021-10-12 16:41 ` [PATCH 12/14] PCI: aardvark: Set PCI Bridge Class Code to PCI Bridge Marek Behún
2021-10-28 18:30 ` Lorenzo Pieralisi
2021-10-28 18:45 ` Pali Rohár
2021-10-28 20:43 ` Bjorn Helgaas
2021-10-12 16:41 ` [PATCH 13/14] PCI: aardvark: Fix support for PCI_BRIDGE_CTL_BUS_RESET on emulated bridge Marek Behún
2021-10-12 16:41 ` [PATCH 14/14] PCI: aardvark: Fix support for PCI_ROM_ADDRESS1 " Marek Behún
2021-10-19 18:36 ` [PATCH 00/14] PCI: aardvark controller fixes BATCH 2 Pali Rohár
2021-10-28 18:33 ` Lorenzo Pieralisi
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20211028110835.GA1846@lpieralisi \
--to=lorenzo.pieralisi@arm.com \
--cc=bhelgaas@google.com \
--cc=kabel@kernel.org \
--cc=linux-pci@vger.kernel.org \
--cc=pali@kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).