Skip to content

cloud-kernel
cloud-kernel

openanolis / cloud-kernel
接近 2 年 前同步成功

通知 170
Star 36
Fork 7
cloud-kernel
cloud-kernel
切换分支/标签
  1. 20 10月, 2012 8 次提交
  3. 12 10月, 2012 1 次提交
    • K
      xen/pv-on-hvm kexec: add quirk for Xen 3.4 and shutdown watches. · cb6b6df1
      由 Konrad Rzeszutek Wilk 提交于 
      The commit 254d1a3f, titled
"xen/pv-on-hvm kexec: shutdown watches from old kernel" assumes that the
XenBus backend can deal with reading of values from:
 "control/platform-feature-xs_reset_watches":

    ... a patch for xenstored is required so that it
    accepts the XS_RESET_WATCHES request from a client (see changeset
    23839:42a45baf037d in xen-unstable.hg). Without the patch for xenstored
    the registration of watches will fail and some features of a PVonHVM
    guest are not available. The guest is still able to boot, but repeated
    kexec boots will fail."

Sadly this is not true when using a Xen 3.4 hypervisor and booting a PVHVM
guest. We end up hanging at:

  err = xenbus_scanf(XBT_NIL, "control",
                        "platform-feature-xs_reset_watches", "%d", &supported);

This can easily be seen with guests hanging at xenbus_init:

NX (Execute Disable) protection: active
SMBIOS 2.4 present.
DMI: Xen HVM domU, BIOS 3.4.0 05/13/2011
Hypervisor detected: Xen HVM
Xen version 3.4.
Xen Platform PCI: I/O protocol version 1
... snip ..
calling  xenbus_init+0x0/0x27e @ 1

Reverting the commit or using the attached patch fixes the issue. This fix
checks whether the hypervisor is older than 4.0 and if so does not try to
perform the read.

Fixes-Oracle-Bug: 14708233
CC: stable@vger.kernel.org
Acked-by: NOlaf Hering <olaf@aepfle.de>
[v2: Added a comment in the source code]
Signed-off-by: NKonrad Rzeszutek Wilk <konrad.wilk@oracle.com>
      cb6b6df1
  5. 03 10月, 2012 2 次提交
  7. 26 9月, 2012 1 次提交
    • K
      xen/pciback: Restore the PCI config space after an FLR. · c341ca45
      由 Konrad Rzeszutek Wilk 提交于 
      When we do an FLR, or D0->D3_hot we may lose the BARs as the
device has turned itself off (and on). This means the device cannot
function unless the pci_restore_state is called - which it is
when the PCI device is unbound from the Xen PCI backend driver.
For PV guests it ends up calling pci_enable_device / pci_enable_msi[x]
which does the proper steps

That however is not happening if a HVM guest is run as QEMU
deals with PCI configuration space. QEMU also requires that the
device be "parked"  under the ownership of a pci-stub driver to
guarantee that the PCI device is not being used. Hence we
follow the same incantation as pci_reset_function does - by
doing an FLR, then restoring the PCI configuration space.

The result of this patch is that when you run lspci, you get
now this:

-       Region 0: [virtual] Memory at fe8c0000 (32-bit, non-prefetchable) [size=128K]
-       Region 1: [virtual] Memory at fe800000 (32-bit, non-prefetchable) [size=512K]
+       Region 0: Memory at fe8c0000 (32-bit, non-prefetchable) [size=128K]
+       Region 1: Memory at fe800000 (32-bit, non-prefetchable) [size=512K]
        Region 2: I/O ports at c000 [size=32]
-       Region 3: [virtual] Memory at fe8e0000 (32-bit, non-prefetchable) [size=16K]
+       Region 3: Memory at fe8e0000 (32-bit, non-prefetchable) [size=16K]

The [virtual] means that lspci read those entries from SysFS but when
it read them from the device it got a different value (0xfffffff).

CC: stable@vger.kernel.org #only for 3.5, 3.6
Signed-off-by: NKonrad Rzeszutek Wilk <konrad.wilk@oracle.com>
      c341ca45
  9. 25 9月, 2012 1 次提交
  11. 21 9月, 2012 1 次提交
    • A
      xen/gndev: Xen backend support for paged out grant targets V4. · c571898f
      由 Andres Lagar-Cavilla 提交于 
      Since Xen-4.2, hvm domains may have portions of their memory paged out. When a
foreign domain (such as dom0) attempts to map these frames, the map will
initially fail. The hypervisor returns a suitable errno, and kicks an
asynchronous page-in operation carried out by a helper. The foreign domain is
expected to retry the mapping operation until it eventually succeeds. The
foreign domain is not put to sleep because itself could be the one running the
pager assist (typical scenario for dom0).

This patch adds support for this mechanism for backend drivers using grant
mapping and copying operations. Specifically, this covers the blkback and
gntdev drivers (which map foreign grants), and the netback driver (which copies
foreign grants).

* Add a retry method for grants that fail with GNTST_eagain (i.e. because the
  target foreign frame is paged out).
* Insert hooks with appropriate wrappers in the aforementioned drivers.

The retry loop is only invoked if the grant operation status is GNTST_eagain.
It guarantees to leave a new status code different from GNTST_eagain. Any other
status code results in identical code execution as before.

The retry loop performs 256 attempts with increasing time intervals through a
32 second period. It uses msleep to yield while waiting for the next retry.

V2 after feedback from David Vrabel:
* Explicit MAX_DELAY instead of wrap-around delay into zero
* Abstract GNTST_eagain check into core grant table code for netback module.

V3 after feedback from Ian Campbell:
* Add placeholder in array of grant table error descriptions for unrelated
  error code we jump over.
* Eliminate single map and retry macro in favor of a generic batch flavor.
* Some renaming.
* Bury most implementation in grant_table.c, cleaner interface.

V4 rebased on top of sync of Xen grant table interface headers.
Signed-off-by: NAndres Lagar-Cavilla <andres@lagarcavilla.org>
Acked-by: NIan Campbell <ian.campbell@citrix.com>
[v5: Fixed whitespace issues]
Signed-off-by: NKonrad Rzeszutek Wilk <konrad.wilk@oracle.com>
      c571898f
  13. 18 9月, 2012 4 次提交
  15. 17 9月, 2012 3 次提交
  17. 11 9月, 2012 1 次提交
  19. 07 9月, 2012 1 次提交
  21. 06 9月, 2012 2 次提交
    • K
      xen/pciback: Fix proper FLR steps. · 80ba77df
      由 Konrad Rzeszutek Wilk 提交于 
      When we do FLR and save PCI config we did it in the wrong order.
The end result was that if a PCI device was unbind from
its driver, then binded to xen-pciback, and then back to its
driver we would get:

> lspci -s 04:00.0
04:00.0 Ethernet controller: Intel Corporation 82574L Gigabit Network Connection
13:42:12 # 4 :~/
> echo "0000:04:00.0" > /sys/bus/pci/drivers/pciback/unbind
> modprobe e1000e
e1000e: Intel(R) PRO/1000 Network Driver - 2.0.0-k
e1000e: Copyright(c) 1999 - 2012 Intel Corporation.
e1000e 0000:04:00.0: Disabling ASPM L0s L1
e1000e 0000:04:00.0: enabling device (0000 -> 0002)
xen: registering gsi 48 triggering 0 polarity 1
Already setup the GSI :48
e1000e 0000:04:00.0: Interrupt Throttling Rate (ints/sec) set to dynamic conservative mode
e1000e: probe of 0000:04:00.0 failed with error -2

This fixes it by first saving the PCI configuration space, then
doing the FLR.
Reported-by: NRen, Yongjie <yongjie.ren@intel.com>
Reported-and-Tested-by: NTobias Geiger <tobias.geiger@vido.info>
Signed-off-by: NKonrad Rzeszutek Wilk <konrad.wilk@oracle.com>
CC: stable@vger.kernel.org
      80ba77df
    • A
      xen/privcmd: add PRIVCMD_MMAPBATCH_V2 ioctl · ceb90fa0
      由 Andres Lagar-Cavilla 提交于 
      PRIVCMD_MMAPBATCH_V2 extends PRIVCMD_MMAPBATCH with an additional
field for reporting the error code for every frame that could not be
mapped.  libxc prefers PRIVCMD_MMAPBATCH_V2 over PRIVCMD_MMAPBATCH.

Also expand PRIVCMD_MMAPBATCH to return appropriate error-encoding top nibble
in the mfn array.
Signed-off-by: NDavid Vrabel <david.vrabel@citrix.com>
Signed-off-by: NAndres Lagar-Cavilla <andres@lagarcavilla.org>
Signed-off-by: NKonrad Rzeszutek Wilk <konrad.wilk@oracle.com>
      ceb90fa0
  23. 05 9月, 2012 1 次提交
  25. 23 8月, 2012 5 次提交
  27. 22 8月, 2012 2 次提交
  29. 17 8月, 2012 1 次提交
  31. 14 9月, 2012 2 次提交
  33. 20 7月, 2012 4 次提交
    • O
      xen PVonHVM: move shared_info to MMIO before kexec · 00e37bdb
      由 Olaf Hering 提交于 
      Currently kexec in a PVonHVM guest fails with a triple fault because the
new kernel overwrites the shared info page. The exact failure depends on
the size of the kernel image. This patch moves the pfn from RAM into
MMIO space before the kexec boot.

The pfn containing the shared_info is located somewhere in RAM. This
will cause trouble if the current kernel is doing a kexec boot into a
new kernel. The new kernel (and its startup code) can not know where the
pfn is, so it can not reserve the page. The hypervisor will continue to
update the pfn, and as a result memory corruption occours in the new
kernel.

One way to work around this issue is to allocate a page in the
xen-platform pci device's BAR memory range. But pci init is done very
late and the shared_info page is already in use very early to read the
pvclock. So moving the pfn from RAM to MMIO is racy because some code
paths on other vcpus could access the pfn during the small   window when
the old pfn is moved to the new pfn. There is even a  small window were
the old pfn is not backed by a mfn, and during that time all reads
return -1.

Because it is not known upfront where the MMIO region is located it can
not be used right from the start in xen_hvm_init_shared_info.

To minimise trouble the move of the pfn is done shortly before kexec.
This does not eliminate the race because all vcpus are still online when
the syscore_ops will be called. But hopefully there is no work pending
at this point in time. Also the syscore_op is run last which reduces the
risk further.
Signed-off-by: NOlaf Hering <olaf@aepfle.de>
Signed-off-by: NKonrad Rzeszutek Wilk <konrad.wilk@oracle.com>
      00e37bdb
    • O
      xen: enable platform-pci only in a Xen guest · 38ad4f4b
      由 Olaf Hering 提交于 
      While debugging kexec issues in a PVonHVM guest I modified
xen_hvm_platform() to return false to disable all PV drivers. This
caused a crash in platform_pci_init() because it expects certain data
structures to be initialized properly.

To avoid such a crash make sure the driver is initialized only if
running in a Xen guest.
Signed-off-by: NOlaf Hering <olaf@aepfle.de>
Signed-off-by: NKonrad Rzeszutek Wilk <konrad.wilk@oracle.com>
      38ad4f4b
    • O
      xen/pv-on-hvm kexec: shutdown watches from old kernel · 254d1a3f
      由 Olaf Hering 提交于 
      Add xs_reset_watches function to shutdown watches from old kernel after
kexec boot.  The old kernel does not unregister all watches in the
shutdown path.  They are still active, the double registration can not
be detected by the new kernel.  When the watches fire, unexpected events
will arrive and the xenwatch thread will crash (jumps to NULL).  An
orderly reboot of a hvm guest will destroy the entire guest with all its
resources (including the watches) before it is rebuilt from scratch, so
the missing unregister is not an issue in that case.

With this change the xenstored is instructed to wipe all active watches
for the guest.  However, a patch for xenstored is required so that it
accepts the XS_RESET_WATCHES request from a client (see changeset
23839:42a45baf037d in xen-unstable.hg). Without the patch for xenstored
the registration of watches will fail and some features of a PVonHVM
guest are not available. The guest is still able to boot, but repeated
kexec boots will fail.
Signed-off-by: NOlaf Hering <olaf@aepfle.de>
Signed-off-by: NKonrad Rzeszutek Wilk <konrad.wilk@oracle.com>
      254d1a3f
    • K
      xen/acpi: Fix potential memory leak. · 17f9b896
      由 Konrad Rzeszutek Wilk 提交于 
      Coverity points out that we do not free in one case the
pr_backup - and sure enough we forgot.

Found by Coverity (CID 401970)
Signed-off-by: NKonrad Rzeszutek Wilk <konrad.wilk@oracle.com>
      17f9b896