From: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
To: Jan Beulich <JBeulich@suse.com>
Cc: xen-devel <xen-devel@lists.xenproject.org>,
Keir Fraser <keir@xen.org>,
Andrew Cooper <andrew.cooper3@citrix.com>
Subject: Re: [PATCH 2/2] x86/vMSI-X: add valid bits for read acceleration
Date: Mon, 23 Mar 2015 15:11:46 -0400 [thread overview]
Message-ID: <20150323191146.GA20509@l.oracle.com> (raw)
In-Reply-To: <550C589D020000780006C2B2@mail.emea.novell.com>
On Fri, Mar 20, 2015 at 04:27:57PM +0000, Jan Beulich wrote:
> Again because Xen doesn't get to see all guest writes, it shouldn't
> serve reads from its cache before having seen a write to the respective
> address.
>
> Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
>
> --- a/xen/arch/x86/hvm/vmsi.c
> +++ b/xen/arch/x86/hvm/vmsi.c
> @@ -153,12 +153,15 @@ struct msixtbl_entry
> /* TODO: resolve the potential race by destruction of pdev */
> struct pci_dev *pdev;
> unsigned long gtable; /* gpa of msix table */
> - unsigned long table_flags[BITS_TO_LONGS(MAX_MSIX_TABLE_ENTRIES)];
> + DECLARE_BITMAP(table_flags, MAX_MSIX_TABLE_ENTRIES);
That seems unrelated to this patch? Perhaps mention the cleanup
part in the commit.
> #define MAX_MSIX_ACC_ENTRIES 3
> unsigned int table_len;
> struct {
> uint32_t msi_ad[3]; /* Shadow of address low, high and data */
> } gentries[MAX_MSIX_ACC_ENTRIES];
> + DECLARE_BITMAP(acc_valid, 3 * MAX_MSIX_ACC_ENTRIES);
> +#define acc_bit(what, ent, slot, idx) \
> + what##_bit((slot) * 3 + (idx), (ent)->acc_valid)
> struct rcu_head rcu;
> };
>
> @@ -233,9 +236,10 @@ static int msixtbl_read(
> if ( offset != PCI_MSIX_ENTRY_VECTOR_CTRL_OFFSET )
> {
> nr_entry = (address - entry->gtable) / PCI_MSIX_ENTRY_SIZE;
> - if ( nr_entry >= MAX_MSIX_ACC_ENTRIES )
> - goto out;
> index = offset / sizeof(uint32_t);
> + if ( nr_entry >= MAX_MSIX_ACC_ENTRIES ||
> + !acc_bit(test, entry, nr_entry, index) )
> + goto out;
> *pval = entry->gentries[nr_entry].msi_ad[index];
> }
> else
> @@ -281,6 +285,7 @@ static int msixtbl_write(struct vcpu *v,
> {
> index = offset / sizeof(uint32_t);
> entry->gentries[nr_entry].msi_ad[index] = val;
> + acc_bit(set, entry, nr_entry, index);
> }
> set_bit(nr_entry, &entry->table_flags);
> goto out;
>
>
>
> x86/vMSI-X: add valid bits for read acceleration
>
> Again because Xen doesn't get to see all guest writes, it shouldn't
> serve reads from its cache before having seen a write to the respective
> address.
>
> Signed-off-by: Jan Beulich <jbeulich@suse.com>
>
> --- a/xen/arch/x86/hvm/vmsi.c
> +++ b/xen/arch/x86/hvm/vmsi.c
> @@ -153,12 +153,15 @@ struct msixtbl_entry
> /* TODO: resolve the potential race by destruction of pdev */
> struct pci_dev *pdev;
> unsigned long gtable; /* gpa of msix table */
> - unsigned long table_flags[BITS_TO_LONGS(MAX_MSIX_TABLE_ENTRIES)];
> + DECLARE_BITMAP(table_flags, MAX_MSIX_TABLE_ENTRIES);
> #define MAX_MSIX_ACC_ENTRIES 3
> unsigned int table_len;
> struct {
> uint32_t msi_ad[3]; /* Shadow of address low, high and data */
> } gentries[MAX_MSIX_ACC_ENTRIES];
> + DECLARE_BITMAP(acc_valid, 3 * MAX_MSIX_ACC_ENTRIES);
> +#define acc_bit(what, ent, slot, idx) \
> + what##_bit((slot) * 3 + (idx), (ent)->acc_valid)
> struct rcu_head rcu;
> };
>
> @@ -233,9 +236,10 @@ static int msixtbl_read(
> if ( offset != PCI_MSIX_ENTRY_VECTOR_CTRL_OFFSET )
> {
> nr_entry = (address - entry->gtable) / PCI_MSIX_ENTRY_SIZE;
> - if ( nr_entry >= MAX_MSIX_ACC_ENTRIES )
> - goto out;
> index = offset / sizeof(uint32_t);
> + if ( nr_entry >= MAX_MSIX_ACC_ENTRIES ||
> + !acc_bit(test, entry, nr_entry, index) )
> + goto out;
> *pval = entry->gentries[nr_entry].msi_ad[index];
> }
> else
> @@ -281,6 +285,7 @@ static int msixtbl_write(struct vcpu *v,
> {
> index = offset / sizeof(uint32_t);
> entry->gentries[nr_entry].msi_ad[index] = val;
> + acc_bit(set, entry, nr_entry, index);
> }
> set_bit(nr_entry, &entry->table_flags);
> goto out;
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xen.org
> http://lists.xen.org/xen-devel
next prev parent reply other threads:[~2015-03-23 19:11 UTC|newest]
Thread overview: 12+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-03-20 15:52 [PATCH 0/2] x86/vMSI-X: table read/write emulation adjustments Jan Beulich
2015-03-20 16:27 ` [PATCH 1/2] x86/vMSI-X: honor all mask requests Jan Beulich
2015-03-23 19:15 ` Konrad Rzeszutek Wilk
2015-03-25 5:40 ` Wu, Feng
2015-03-25 5:49 ` Wu, Feng
2015-03-25 10:36 ` Jan Beulich
2015-04-14 14:17 ` Andrew Cooper
2015-03-20 16:27 ` [PATCH 2/2] x86/vMSI-X: add valid bits for read acceleration Jan Beulich
2015-03-23 19:11 ` Konrad Rzeszutek Wilk [this message]
2015-03-24 7:49 ` Jan Beulich
2015-04-14 14:37 ` Andrew Cooper
2015-04-14 10:11 ` Ping: [PATCH 0/2] x86/vMSI-X: table read/write emulation adjustments 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=20150323191146.GA20509@l.oracle.com \
--to=konrad.wilk@oracle.com \
--cc=JBeulich@suse.com \
--cc=andrew.cooper3@citrix.com \
--cc=keir@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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.