kvm.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Jan Kiszka <jan.kiszka@siemens.com>
To: Alex Williamson <alex.williamson@redhat.com>
Cc: Avi Kivity <avi@redhat.com>,
	Marcelo Tosatti <mtosatti@redhat.com>,
	"kvm@vger.kernel.org" <kvm@vger.kernel.org>
Subject: [PATCH v3 4/6] pci-assign: Properly handle more overlapping accesses
Date: Mon, 02 May 2011 12:29:39 +0200	[thread overview]
Message-ID: <4DBE8793.3090505@siemens.com> (raw)
In-Reply-To: <1304109696.3266.19.camel@x201>

On 2011-04-29 22:41, Alex Williamson wrote:
> On Fri, 2011-04-29 at 11:05 +0200, Jan Kiszka wrote:
>> Ensure that accesses exceeding PCI_CAPABILITY_LIST and
>> PCI_INTERRUPT_LINE+PIN hit the real device in areas we do not
>> virtualize. Again, we do not optimize these checks and accesses a lot,
>> they are considered to be slow paths.
>>
>> Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
>> ---
>>  hw/device-assignment.c |   34 +++++++++++++++++++++++++++++-----
>>  1 files changed, 29 insertions(+), 5 deletions(-)
>>
>> diff --git a/hw/device-assignment.c b/hw/device-assignment.c
>> index cea072e..37c77e3 100644
>> --- a/hw/device-assignment.c
>> +++ b/hw/device-assignment.c
>> @@ -442,7 +442,29 @@ static void assigned_dev_pci_write_config(PCIDevice *d, uint32_t address,
>>          ranges_overlap(address, len, PCI_INTERRUPT_LINE, 2)) {
>>          /* used for update-mappings (BAR emulation) */
>>          pci_default_write_config(d, address, val, len);
>> -        return;
>> +
>> +        /* Ensure that writes to overlapping areas we don't virtualize still
>> +         * hit the device. */
>> +        switch (address) {
>> +        case PCI_CAPABILITY_LIST:
>> +            if (len > 1) {
>> +                len -= 1;
>> +                address += 1;
>> +                val >>= 8;
>> +                break; /* continue writing to the device */
>> +            }
>> +            return;
>> +        case PCI_INTERRUPT_LINE:
>> +            if (len > 2) {
>> +                len -= 2;
>> +                address += 2;
>> +                val >>= 16;
>> +                break; /* continue writing to the device */
>> +            }
>> +            return;
>> +        default:
>> +            return;
>> +        }
>>      }
> 
> It seems like we could be more symmetric with the below read.  Maybe
> something like:
> 
> if (ranges_overlap(address, len, PCI_BASE_ADDRESS_0, 24) ||
>     ranges_overlap(address, len, PCI_ROM_ADDRESS, 4)) {
>     pci_default_write_config(d, address, val, len);
>     return;
> } else if (ranges_overlap(address, len, PCI_CAPABILITY_LIST, 1) ||
>            ranges_overlap(address, len, PCI_INTERRUPT_LINE, 2)) {
>     uint32_t real_val =  assigned_dev_pci_read(d, address, len);
>     pci_default_write_config(d, address, val, len);
>     val = merge_bits(val, real_val, address, len, PCI_CAPABILITY_LIST, 0xff);
>     val = merge_bits(val, real_val, address, len, PCI_INTERRUPT_LINE, 0xffff);
> }
> 
> We might write out to real hardware when we could avoid it, but I don't
> think it matters.  Thanks,
> 

Yep, makes sense. Find such a version below.

Thanks,
Jan


------8<-------

Ensure that accesses exceeding PCI_CAPABILITY_LIST and
PCI_INTERRUPT_LINE+PIN hit the real device in areas we do not
virtualize. Again, we do not optimize these checks and accesses a lot,
they are considered to be slow paths.

Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
---
 hw/device-assignment.c |   34 ++++++++++++++++++++++++++--------
 1 files changed, 26 insertions(+), 8 deletions(-)

diff --git a/hw/device-assignment.c b/hw/device-assignment.c
index cea072e..8e95730 100644
--- a/hw/device-assignment.c
+++ b/hw/device-assignment.c
@@ -438,11 +438,22 @@ static void assigned_dev_pci_write_config(PCIDevice *d, uint32_t address,
      *  - interrupt line & pin
      */
     if (ranges_overlap(address, len, PCI_BASE_ADDRESS_0, 24) ||
-        ranges_overlap(address, len, PCI_ROM_ADDRESS, 5) ||
-        ranges_overlap(address, len, PCI_INTERRUPT_LINE, 2)) {
-        /* used for update-mappings (BAR emulation) */
+        ranges_overlap(address, len, PCI_ROM_ADDRESS, 4)) {
         pci_default_write_config(d, address, val, len);
         return;
+    } else if (ranges_overlap(address, len, PCI_CAPABILITY_LIST, 1) ||
+               ranges_overlap(address, len, PCI_INTERRUPT_LINE, 2)) {
+        uint32_t real_val;
+
+        pci_default_write_config(d, address, val, len);
+
+        /* Ensure that writes to overlapping areas we don't virtualize still
+         * hit the device. */
+        real_val = assigned_dev_pci_read(d, address, len);
+        val = merge_bits(val, real_val, address, len,
+                         PCI_CAPABILITY_LIST, 0xff);
+        val = merge_bits(val, real_val, address, len,
+                         PCI_INTERRUPT_LINE, 0xffff);
     }
 
     DEBUG("NON BAR (%x.%x): address=%04x val=0x%08x len=%d\n",
@@ -467,7 +478,7 @@ again:
 static uint32_t assigned_dev_pci_read_config(PCIDevice *d, uint32_t address,
                                              int len)
 {
-    uint32_t val = 0;
+    uint32_t val = 0, virt_val;
     int fd;
     ssize_t ret;
     AssignedDevice *pci_dev = DO_UPCAST(AssignedDevice, dev, d);
@@ -483,13 +494,11 @@ static uint32_t assigned_dev_pci_read_config(PCIDevice *d, uint32_t address,
      * Catch access to
      *  - vendor & device ID
      *  - base address registers
-     *  - ROM base address & capability pointer
-     *  - interrupt line & pin
+     *  - ROM base address
      */
     if (ranges_overlap(address, len, PCI_VENDOR_ID, 4) ||
         ranges_overlap(address, len, PCI_BASE_ADDRESS_0, 24) ||
-        ranges_overlap(address, len, PCI_ROM_ADDRESS, 5) ||
-        ranges_overlap(address, len, PCI_INTERRUPT_LINE, 2)) {
+        ranges_overlap(address, len, PCI_ROM_ADDRESS, 4)) {
         val = pci_default_read_config(d, address, len);
         DEBUG("(%x.%x): address=%04x val=0x%08x len=%d\n",
               (d->devfn >> 3) & 0x1F, (d->devfn & 0x7), address, val, len);
@@ -523,6 +532,15 @@ do_log:
                          address, len, PCI_COMMAND, 0xffff);
     }
 
+    /*
+     * Merge bits from virtualized
+     *  - capability pointer
+     *  - interrupt line & pin
+     */
+    virt_val = pci_default_read_config(d, address, len);
+    val = merge_bits(val, virt_val, address, len, PCI_CAPABILITY_LIST, 0xff);
+    val = merge_bits(val, virt_val, address, len, PCI_INTERRUPT_LINE, 0xffff);
+
     if (!pci_dev->cap.available) {
         /* kill the special capabilities */
         if (address == PCI_COMMAND && len == 4) {
-- 
1.7.1

  reply	other threads:[~2011-05-02 10:29 UTC|newest]

Thread overview: 12+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2011-04-29  9:05 [PATCH v2 0/6] qemu-kvm: pci-assign: Some more cleanups Jan Kiszka
2011-04-29  9:05 ` [PATCH v2 1/6] pci-assign: Clean up assigned_dev_pci_read/write_config Jan Kiszka
2011-04-29  9:05 ` [PATCH v2 2/6] pci-assign: Move merge_bits Jan Kiszka
2011-04-29  9:05 ` [PATCH v2 3/6] pci-assign: Fix dword read at PCI_COMMAND Jan Kiszka
2011-04-29  9:05 ` [PATCH v2 4/6] pci-assign: Properly handle more overlapping accesses Jan Kiszka
2011-04-29 20:41   ` Alex Williamson
2011-05-02 10:29     ` Jan Kiszka [this message]
2011-05-02 14:09       ` [PATCH v3 " Alex Williamson
2011-04-29  9:05 ` [PATCH v2 5/6] pci-assign: Remove suspicious hunk from assigned_dev_pci_read_config Jan Kiszka
2011-04-29  9:05 ` [PATCH v2 6/6] pci-assign: Convert need_emulate_cmd into a bitmask Jan Kiszka
2011-05-02 14:11 ` [PATCH v2 0/6] qemu-kvm: pci-assign: Some more cleanups Alex Williamson
2011-05-03 19:17 ` Marcelo Tosatti

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=4DBE8793.3090505@siemens.com \
    --to=jan.kiszka@siemens.com \
    --cc=alex.williamson@redhat.com \
    --cc=avi@redhat.com \
    --cc=kvm@vger.kernel.org \
    --cc=mtosatti@redhat.com \
    /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).