[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20200203143549.GG83200@debian-boqun.qqnc3lrjykvubdpftowmye0fmh.lx.internal.cloudapp.net>
Date: Mon, 3 Feb 2020 22:35:49 +0800
From: Boqun Feng <boqun.feng@...il.com>
To: Andrew Murray <amurray@...goodpenguin.co.uk>
Cc: linux-pci@...r.kernel.org, linux-hyperv@...r.kernel.org,
linux-kernel@...r.kernel.org, linux-arm-kernel@...ts.infradead.org,
Michael Kelley <mikelley@...rosoft.com>,
"K. Y. Srinivasan" <kys@...rosoft.com>,
Haiyang Zhang <haiyangz@...rosoft.com>,
Stephen Hemminger <sthemmin@...rosoft.com>,
Sasha Levin <sashal@...nel.org>,
Thomas Gleixner <tglx@...utronix.de>,
Ingo Molnar <mingo@...hat.com>, Borislav Petkov <bp@...en8.de>,
"H. Peter Anvin" <hpa@...or.com>, x86@...nel.org,
Lorenzo Pieralisi <lorenzo.pieralisi@....com>,
Andrew Murray <andrew.murray@....com>,
Bjorn Helgaas <bhelgaas@...gle.com>
Subject: Re: [PATCH v2 3/3] PCI: hv: Introduce hv_msi_entry
On Mon, Feb 03, 2020 at 09:51:40AM +0000, Andrew Murray wrote:
> On Mon, Feb 03, 2020 at 01:03:13PM +0800, Boqun Feng wrote:
> > Add a new structure (hv_msi_entry), which is also defined int tlfs, to
>
> s/int/in the/ ?
>
Good catch, will fix.
> > describe the msi entry for HVCALL_RETARGET_INTERRUPT. The structure is
> > needed because its layout may be different from architecture to
> > architecture.
> >
> > Also add a new generic interface hv_set_msi_address_from_desc() to allow
> > different archs to set the msi address from msi_desc.
> >
> > No functional change, only preparation for the future support of virtual
> > PCI on non-x86 architectures.
> >
> > Signed-off-by: Boqun Feng (Microsoft) <boqun.feng@...il.com>
> > ---
> > arch/x86/include/asm/hyperv-tlfs.h | 11 +++++++++--
> > arch/x86/include/asm/mshyperv.h | 5 +++++
> > drivers/pci/controller/pci-hyperv.c | 4 ++--
> > 3 files changed, 16 insertions(+), 4 deletions(-)
> >
> > diff --git a/arch/x86/include/asm/hyperv-tlfs.h b/arch/x86/include/asm/hyperv-tlfs.h
> > index 4a76e442481a..953b3ad38746 100644
> > --- a/arch/x86/include/asm/hyperv-tlfs.h
> > +++ b/arch/x86/include/asm/hyperv-tlfs.h
> > @@ -912,11 +912,18 @@ struct hv_partition_assist_pg {
> > u32 tlb_lock_count;
> > };
> >
> > +union hv_msi_entry {
> > + u64 as_uint64;
> > + struct {
> > + u32 address;
> > + u32 data;
> > + } __packed;
> > +};
> > +
> > struct hv_interrupt_entry {
> > u32 source; /* 1 for MSI(-X) */
> > u32 reserved1;
> > - u32 address;
> > - u32 data;
> > + union hv_msi_entry msi_entry;
> > } __packed;
> >
> > /*
> > diff --git a/arch/x86/include/asm/mshyperv.h b/arch/x86/include/asm/mshyperv.h
> > index 6b79515abb82..3bdaa3b6e68f 100644
> > --- a/arch/x86/include/asm/mshyperv.h
> > +++ b/arch/x86/include/asm/mshyperv.h
> > @@ -240,6 +240,11 @@ bool hv_vcpu_is_preempted(int vcpu);
> > static inline void hv_apic_init(void) {}
> > #endif
> >
> > +#define hv_set_msi_address_from_desc(msi_entry, msi_desc) \
> > +do { \
> > + (msi_entry)->address = (msi_desc)->msg.address_lo; \
> > +} while (0)
>
> Given that this is a single statement, is there really a need for the do ; while(0) ?
>
I choose to use do ; while (0) because I don't want code like the
following to be able to compile:
hv_set_msi_address_from_desc(...) /* semicolon is missing */
a = b;
But now think more about this, I think it's probably better to define
this as a function..
>
> > +
> > #else /* CONFIG_HYPERV */
> > static inline void hyperv_init(void) {}
> > static inline void hyperv_setup_mmu_ops(void) {}
> > diff --git a/drivers/pci/controller/pci-hyperv.c b/drivers/pci/controller/pci-hyperv.c
> > index 0d9b74503577..2240f2b3643e 100644
> > --- a/drivers/pci/controller/pci-hyperv.c
> > +++ b/drivers/pci/controller/pci-hyperv.c
> > @@ -1170,8 +1170,8 @@ static void hv_irq_unmask(struct irq_data *data)
> > memset(params, 0, sizeof(*params));
> > params->partition_id = HV_PARTITION_ID_SELF;
> > params->int_entry.source = 1; /* MSI(-X) */
> > - params->int_entry.address = msi_desc->msg.address_lo;
> > - params->int_entry.data = msi_desc->msg.data;
> > + hv_set_msi_address_from_desc(¶ms->int_entry.msi_entry, msi_desc);
> > + params->int_entry.msi_entry.data = msi_desc->msg.data;
>
> If the layout may differ, then don't we also need a wrapper for data?
>
On x86 hv_msi_entry is:
{
u32 address;
u32 data;
}
and on ARM64 it is:
{
u64 address;
u32 data;
u32 reserved;
}
So currently, setting msi_entry.data doesn't need a wrapper for
different archs. But now you mention it, probably a better way is to
provide a wrapper hv_set_msi_entry_from_desc(), which sets both address
and data instead of hv_set_msi_address_from_desc().
Thanks for looking into the whole patchset!
Regards,
Boqun
> Thanks,
>
> Andrew Murray
>
> > params->device_id = (hbus->hdev->dev_instance.b[5] << 24) |
> > (hbus->hdev->dev_instance.b[4] << 16) |
> > (hbus->hdev->dev_instance.b[7] << 8) |
> > --
> > 2.24.1
> >
Powered by blists - more mailing lists