From: Rahul Singh <Rahul.Singh@arm.com>
To: Jan Beulich <jbeulich@suse.com>
Cc: "Roger Pau Monné" <roger.pau@citrix.com>,
xen-devel <xen-devel@lists.xenproject.org>,
"Bertrand Marquis" <Bertrand.Marquis@arm.com>,
"Andrew Cooper" <andrew.cooper3@citrix.com>,
"Wei Liu" <wl@xen.org>,
"George Dunlap" <george.dunlap@citrix.com>,
"Ian Jackson" <iwj@xenproject.org>,
"Julien Grall" <julien@xen.org>,
"Stefano Stabellini" <sstabellini@kernel.org>,
"Paul Durrant" <paul@xen.org>,
"Volodymyr Babchuk" <Volodymyr_Babchuk@epam.com>,
"Daniel De Graaf" <dgdegra@tycho.nsa.gov>
Subject: Re: [PATCH v2] xen/pci: Refactor PCI MSI interrupts related code
Date: Mon, 12 Apr 2021 18:26:28 +0000 [thread overview]
Message-ID: <9518F24F-EDA3-4C0E-B2A6-DB4A0F868EC3@arm.com> (raw)
In-Reply-To: <7f7383a5-a678-0dbd-cf9f-8a9239f6d391@suse.com>
Hi Jan,
> On 12 Apr 2021, at 12:28 pm, Jan Beulich <jbeulich@suse.com> wrote:
>
> On 12.04.2021 12:49, Roger Pau Monné wrote:
>> On Fri, Apr 09, 2021 at 05:00:41PM +0100, Rahul Singh wrote:
>>> --- a/xen/arch/x86/msi.c
>>> +++ b/xen/arch/x86/msi.c
>>> @@ -1411,6 +1411,70 @@ void __init early_msi_init(void)
>>> return;
>>> }
>>>
>>> +int alloc_pci_msi(struct pci_dev *pdev)
>>
>> I would rather name it pci_msi_init...
>
> Or maybe pdev_msi_init(), as pci_msi_init() looks more like a one-
> time, global init function?
>
Ok. I will use pdev_msi_init().
>>> +void free_pci_msi(struct pci_dev *pdev)
>>
>> ...and pci_msi_free.
>
> The counterpart of "init" really would be "deinit", IOW I'd like to
> ask for either alloc/free or init/deinit.
Ok. I will use pdev_msi_deinit().
>
>>> +{
>>> + xfree(pdev->msix);
>
> Could this maybe become XFREE() at this occasion?
Ok.
>
>>> +int pci_assign_msix_init(struct domain *d, struct pci_dev *pdev)
>>
>> This kind of a confusing name - what about pci_msix_assign?
>>
>>> +{
>>> + int rc;
>>> +
>>> + if ( pdev->msix )
>
> Wouldn't this check better live in the sole caller?
Ok.
>
>
>>> +void dump_pci_msi(struct pci_dev *pdev)
>
> pdev_dump_msi() or some such?
>
> Also - const here and ...
Ok.
>
>>> +{
>>> + struct msi_desc *msi;
>
> ... here please, while you already move this code.
Ok.
>
>>> + list_for_each_entry ( msi, &pdev->msi_list, list )
>>> + printk("%d ", msi->irq);
>>> +}
>>
>> I wonder, those those function really want to be in a x86 specific
>> file? There's nothing x86 specific about them AFAICT.
>>
>> Would it make sense to create a separate msi-intercept.c file with
>> those that gets included when CONFIG_PCI_MSI_INTERCEPT?
>
> +1
>
>>> @@ -1303,12 +1279,15 @@ static int __init setup_dump_pcidevs(void)
>>> }
>>> __initcall(setup_dump_pcidevs);
>>>
>>> +
>>> +#ifdef CONFIG_PCI_MSI_INTERCEPT
>
> No double blank lines please.
Ok.
>
>>> int iommu_update_ire_from_msi(
>>> struct msi_desc *msi_desc, struct msi_msg *msg)
>>> {
>>> return iommu_intremap
>>> ? iommu_call(&iommu_ops, update_ire_from_msi, msi_desc, msg) : 0;
>>> }
>>> +#endif
>>
>> This is not exactly related to MSI intercepts, the IOMMU interrupt
>> remapping table will also be used for interrupt entries for devices
>> being used by Xen directly, where no intercept is required.
>>
>> And then you also want to gate the hook from iommu_ops itself with
>> CONFIG_PCI_MSI_INTERCEPT, if we want to got this route.
>
> I think the two aspects of MSI should be kept separate.
Ok. I will split the patch in two patches.
>
>>> --- a/xen/drivers/pci/Kconfig
>>> +++ b/xen/drivers/pci/Kconfig
>>> @@ -1,3 +1,7 @@
>>>
>>> config HAS_PCI
>>> bool
>>> +
>>> +config PCI_MSI_INTERCEPT
>>> + def_bool y
>>> + depends on X86 && HAS_PCI
>
> Depending on HAS_PCI is fine (albeit not strictly needed afaict),
> but this shouldn't have a default (like HAS_PCI doesn't) and
> instead be selected by x86.
Ok.
>
>>> --- a/xen/drivers/vpci/vpci.c
>>> +++ b/xen/drivers/vpci/vpci.c
>>> @@ -48,8 +48,7 @@ void vpci_remove_device(struct pci_dev *pdev)
>>> xfree(r);
>>> }
>>> spin_unlock(&pdev->vpci->lock);
>>> - xfree(pdev->vpci->msix);
>>> - xfree(pdev->vpci->msi);
>>> + free_vpci_msi(pdev);
>
> I don't think the function needs to be passed a pdev, and ...
Ok.
>>> xfree(pdev->vpci);
>>> pdev->vpci = NULL;
>
> ... it would only be consistent with this if pdev->vpci was passed
> instead.
OK.
>
>>> --- /dev/null
>>> +++ b/xen/include/asm-arm/msi.h
>>> @@ -0,0 +1,44 @@
>>> +/*
>>> + * Copyright (C) 2021 Arm Ltd.
>>> + *
>>> + * This program is free software; you can redistribute it and/or modify
>>> + * it under the terms of the GNU General Public License version 2 as
>>> + * published by the Free Software Foundation.
>>> + *
>>> + * This program is distributed in the hope that it will be useful,
>>> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
>>> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
>>> + * GNU General Public License for more details.
>>> + *
>>> + * You should have received a copy of the GNU General Public License
>>> + * along with this program. If not, see <http://www.gnu.org/licenses/>.
>>> + */
>>> +
>>> +#ifndef __ASM_MSI_H_
>>> +#define __ASM_MSI_H_
>>> +
>>> +static inline int alloc_pci_msi(struct pci_dev *pdev)
>>> +{
>>> + return 0;
>>> +}
>>> +
>>> +static inline int pci_assign_msix_init(struct domain *d, struct pci_dev *pdev)
>>> +{
>>> + return 0;
>>> +}
>>> +
>>> +static inline void dump_pci_msi(struct pci_dev *pdev) {}
>>> +static inline void free_pci_msi(struct pci_dev *pdev) {}
>>> +static inline void pci_cleanup_msi(struct pci_dev *pdev) {}
>>> +
>>> +#endif /* __ASM_MSI_H */
>>> +
>>> +/*
>>> + * Local variables:
>>> + * mode: C
>>> + * c-file-style: "BSD"
>>> + * c-basic-offset: 4
>>> + * tab-width: 4
>>> + * indent-tabs-mode: nil
>>> + * End:
>>> + */
>>
>> Should you instead have a non-arch specific file with those dummy
>> helpers that get used when !CONFIG_PCI_MSI_INTERCEPT?
>
> +1
>
>>> --- a/xen/include/asm-x86/msi.h
>>> +++ b/xen/include/asm-x86/msi.h
>>> @@ -252,5 +252,9 @@ void unmask_msi_irq(struct irq_desc *);
>>> void guest_mask_msi_irq(struct irq_desc *, bool mask);
>>> void ack_nonmaskable_msi_irq(struct irq_desc *);
>>> void set_msi_affinity(struct irq_desc *, const cpumask_t *);
>>> +int alloc_pci_msi(struct pci_dev *pdev);
>>> +void free_pci_msi(struct pci_dev *pdev);
>>> +void dump_pci_msi(struct pci_dev *pdev);
>>> +int pci_assign_msix_init(struct domain *d, struct pci_dev *pdev);
>
> These should then live in the other "half" of that header.
Ok Yes noted
>
>>> --- a/xen/include/xen/pci.h
>>> +++ b/xen/include/xen/pci.h
>>> @@ -79,10 +79,6 @@ struct pci_dev {
>>> struct list_head alldevs_list;
>>> struct list_head domain_list;
>>>
>>> - struct list_head msi_list;
>>> -
>>> - struct arch_msix *msix;
>>> -
>>> struct domain *domain;
>>>
>>> const union {
>>> @@ -94,7 +90,14 @@ struct pci_dev {
>>> pci_sbdf_t sbdf;
>>> };
>>>
>>> +#ifdef CONFIG_PCI_MSI_INTERCEPT
>>> + struct list_head msi_list;
>>> +
>>> + struct arch_msix *msix;
>>> +
>>> uint8_t msi_maxvec;
>>> +#endif
>>> +
>>> uint8_t phantom_stride;
>
> Like Roger also said for struct vpci, it's not clear this is worth
> it. And while you may have paid attention (and there simply is no
> better arrangement) I'd also like to point out that such field
> movement should be done while keeping padding hole sizes in mind.
Ok.
>
>>> @@ -174,7 +177,7 @@ int __must_check vpci_msix_arch_disable_entry(struct vpci_msix_entry *entry,
>>> const struct pci_dev *pdev);
>>> void vpci_msix_arch_init_entry(struct vpci_msix_entry *entry);
>>> int vpci_msix_arch_print(const struct vpci_msix *msix);
>>> -
>>> +int remove_msix_regions(const struct pci_dev *pdev, struct rangeset *mem);
>>> /*
>>> * Helper functions to fetch MSIX related data. They are used by both the
>>> * emulated MSIX code and the BAR handlers.
>
> Please don't remove blank lines like this one, unless you actually
> see a reason.
>
Ok.
Regards,
Rahul
> Jan
next prev parent reply other threads:[~2021-04-12 18:27 UTC|newest]
Thread overview: 28+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-04-09 16:00 [PATCH v2] xen/pci: Refactor PCI MSI interrupts related code Rahul Singh
2021-04-12 10:49 ` Roger Pau Monné
2021-04-12 11:28 ` Jan Beulich
2021-04-12 18:26 ` Rahul Singh [this message]
2021-04-12 16:28 ` Rahul Singh
2021-04-13 17:12 ` Julien Grall
2021-04-14 7:08 ` Jan Beulich
2021-04-14 8:28 ` Roger Pau Monné
2021-04-14 8:47 ` Jan Beulich
2021-04-14 8:28 ` Julien Grall
2021-04-14 8:05 ` Roger Pau Monné
2021-04-14 8:49 ` Julien Grall
2021-04-15 13:26 ` Roger Pau Monné
2021-04-15 13:31 ` Julien Grall
2021-04-19 7:16 ` Rahul Singh
2021-04-19 8:40 ` Roger Pau Monné
2021-04-19 11:16 ` Jan Beulich
2021-04-19 11:54 ` Julien Grall
2021-04-19 12:33 ` Jan Beulich
2021-04-20 13:45 ` Rahul Singh
2021-04-20 15:36 ` Jan Beulich
2021-04-21 8:07 ` Rahul Singh
2021-04-21 8:16 ` Roger Pau Monné
2021-04-21 9:15 ` Rahul Singh
2021-04-21 9:33 ` Jan Beulich
2021-04-21 11:55 ` Rahul Singh
2021-04-21 9:32 ` Jan Beulich
2021-04-21 9:01 ` Jan Beulich
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=9518F24F-EDA3-4C0E-B2A6-DB4A0F868EC3@arm.com \
--to=rahul.singh@arm.com \
--cc=Bertrand.Marquis@arm.com \
--cc=Volodymyr_Babchuk@epam.com \
--cc=andrew.cooper3@citrix.com \
--cc=dgdegra@tycho.nsa.gov \
--cc=george.dunlap@citrix.com \
--cc=iwj@xenproject.org \
--cc=jbeulich@suse.com \
--cc=julien@xen.org \
--cc=paul@xen.org \
--cc=roger.pau@citrix.com \
--cc=sstabellini@kernel.org \
--cc=wl@xen.org \
--cc=xen-devel@lists.xenproject.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).