Skip to content

Q
qemu

openeuler / qemu

通知 10
Star 0
Fork 0
Q
qemu

体验新版 GitCode，发现更多精彩内容 >>

切换分支/标签
  1. 19 5月, 2016 1 次提交
  3. 10 2月, 2016 4 次提交
  5. 07 2月, 2016 1 次提交
    • S
      fix MSI injection on Xen · 428c3ece
      由 Stefano Stabellini 提交于 
      On Xen MSIs can be remapped into pirqs, which are a type of event
channels. It's mostly for the benefit of PCI passthrough devices, to
avoid the overhead of interacting with the emulated lapic.

However remapping interrupts and MSIs is also supported for emulated
devices, such as the e1000 and virtio-net.

When an interrupt or an MSI is remapped into a pirq, masking and
unmasking is done by masking and unmasking the event channel. The
masking bit on the PCI config space or MSI-X table should be ignored,
but it isn't at the moment.

As a consequence emulated devices which use MSI or MSI-X, such as
virtio-net, don't work properly (the guest doesn't receive any
notifications). The mechanism was working properly when xen_apic was
introduced, but I haven't narrowed down which commit in particular is
causing the regression.

Fix the issue by ignoring the masking bit for MSI and MSI-X which have
been remapped into pirqs.
Signed-off-by: NStefano Stabellini <stefano.stabellini@eu.citrix.com>
Reviewed-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>
      428c3ece
  7. 29 1月, 2016 1 次提交
    • P
      xen: Clean up includes · 21cbfe5f
      由 Peter Maydell 提交于 
      Clean up includes so that osdep.h is included first and headers
which it implies are not included manually.

This commit was created with scripts/clean-includes.
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>
Message-id: 1453832250-766-14-git-send-email-peter.maydell@linaro.org
      21cbfe5f
  9. 27 1月, 2016 3 次提交
    • I
      xen: Switch uses of xc_map_foreign_{pages,bulk} to use libxenforeignmemory API. · e0cb42ae
      由 Ian Campbell 提交于 
      In Xen 4.7 we are refactoring parts libxenctrl into a number of
separate libraries which will provide backward and forward API and ABI
compatiblity.

One such library will be libxenforeignmemory which provides access to
privileged foreign mappings and which will provide an interface
equivalent to xc_map_foreign_{pages,bulk}.

The new xenforeignmemory_map() function behaves like
xc_map_foreign_pages() when the err argument is NULL and like
xc_map_foreign_bulk() when err is non-NULL, which maps into the shim
here onto checking err == NULL and calling the appropriate old
function.

Note that xenforeignmemory_map() takes the number of pages before the
arrays themselves, in order to support potentially future use of
variable-length-arrays in the prototype (in the future, when Xen's
baseline toolchain requirements are new enough to ensure VLAs are
supported).

In preparation for adding support for libxenforeignmemory add support
to the <=4.0 and <=4.6 compat code in xen_common.h to allow us to
switch to using the new API. These shims will disappear for versions
of Xen which include libxenforeignmemory.

Since libxenforeignmemory will have its own handle type but for <= 4.6
the functionality is provided by using a libxenctrl handle we
introduce a new global xen_fmem alongside the existing xen_xc. In fact
we make xen_fmem a pointer to the existing xen_xc, which then works
correctly with both <=4.0 (xc handle is an int) and <=4.6 (xc handle
is a pointer). In the latter case xen_fmem is actually a double
indirect pointer, but it all falls out in the wash.

Unlike libxenctrl libxenforeignmemory has an explicit unmap function,
rather than just specifying that munmap should be used, so the unmap
paths are updated to use xenforeignmemory_unmap, which is a shim for
munmap on these versions of xen. The mappings in xen-hvm.c do not
appear to be unmapped (which makes sense for a qemu-dm process)

In fb_disconnect this results in a change from simply mmap over the
existing mapping (with an implicit munmap) to expliclty unmapping with
xenforeignmemory_unmap and then mapping the required anonymous memory
in the same hole. I don't think this is a problem since any other
thread which was racily touching this region would already be running
the risk of hitting the mapping halfway through the call. If this is
thought to be a problem then we could consider adding an extra API to
the libxenforeignmemory interface to replace a foreign mapping with
anonymous shared memory, but I'd prefer not to.
Signed-off-by: NIan Campbell <ian.campbell@citrix.com>
Reviewed-by: NStefano Stabellini <stefano.stabellini@eu.citrix.com>
Signed-off-by: NStefano Stabellini <stefano.stabellini@eu.citrix.com>
      e0cb42ae
    • I
      xen: Switch uses of xc_map_foreign_range into xc_map_foreign_pages · 9ed257d1
      由 Ian Campbell 提交于 
      In Xen 4.7 we are refactoring parts libxenctrl into a number of
separate libraries which will provide backward and forward API and ABI
compatiblity.

One such library will be libxenforeignmemory which provides access to
privileged foreign mappings and which will provide an interface
equivalent to xc_map_foreign_{pages,bulk}.

In preparation for this switch all uses of xc_map_foreign_range to
xc_map_foreign_pages. This is trivial because size was always
XC_PAGE_SIZE so the necessary adjustments are trivial:

  * Pass &mfn (an array of length 1) instead of mfn. The function
    takes a pointer to const, so there is no possibily of mfn changing
    due to this change.
  * Pass nr_pages=1 instead of size=XC_PAGE_SIZE

There is one wrinkle in xen_console.c:con_initialise() where
con->ring_ref is an int but can in some code paths (when !xendev->dev)
be treated as an mfn. I think this is an existing latent truncation
hazard on platforms where xen_pfn_t is 64-bit and int is 32-bit (e.g.
amd64, both arm* variants). I'm unsure under what circumstances
xendev->dev can be NULL or if anything elsewhere ensures the value
fits into an int. For now I just use a temporary xen_pfn_t to in
effect upcast the pointer from int* to xen_pfn_t*.

In xenfb.c:common_bind we now explicitly launder the mfn into a
xen_pfn_t, so it has the correct type to be passed to
xc_map_foreign_pages and doesn't provoke warnings on 32-bit x86.
Signed-off-by: NIan Campbell <ian.campbell@citrix.com>
Reviewed-by: NStefano Stabellini <stefano.stabellini@eu.citrix.com>
Signed-off-by: NStefano Stabellini <stefano.stabellini@eu.citrix.com>
      9ed257d1
    • I
      xen: Switch to libxenevtchn interface for compat shims. · a2db2a1e
      由 Ian Campbell 提交于 
      In Xen 4.7 we are refactoring parts libxenctrl into a number of
separate libraries which will provide backward and forward API and ABI
compatiblity.

One such library will be libxenevtchn which provides access to event
channels.

In preparation for this switch the compatibility layer in xen_common.h
(which support building with older versions of Xen) to use what will
be the new library API. This means that the evtchn shim will disappear
for versions of Xen which include libxenevtchn.

To simplify things for the <= 4.0.0 support we wrap the int fd in a
malloc(sizeof int) such that the handle is always a pointer. This
leads to less typedef headaches and the need for
XC_HANDLER_INITIAL_VALUE etc for these interfaces.

Note that this patch does not add any support for actually using
libxenevtchn, it just adjusts the existing shims.

Note that xc_evtchn_alloc_unbound functionality remains in libxenctrl,
since that functionality is not exposed by /dev/xen/evtchn.
Signed-off-by: NIan Campbell <ian.campbell@citrix.com>
Reviewed-by: NStefano Stabellini <stefano.stabellini@eu.citrix.com>
Signed-off-by: NStefano Stabellini <stefano.stabellini@eu.citrix.com>
      a2db2a1e
  11. 15 1月, 2016 2 次提交
  13. 13 1月, 2016 1 次提交
  15. 18 9月, 2015 1 次提交
    • M
      Fix bad error handling after memory_region_init_ram() · f8ed85ac
      由 Markus Armbruster 提交于 
      Symptom:

    $ qemu-system-x86_64 -m 10000000
    Unexpected error in ram_block_add() at /work/armbru/qemu/exec.c:1456:
    upstream-qemu: cannot set up guest memory 'pc.ram': Cannot allocate memory
    Aborted (core dumped)

Root cause: commit ef701d7b screwed up handling of out-of-memory
conditions.  Before the commit, we report the error and exit(1), in
one place, ram_block_add().  The commit lifts the error handling up
the call chain some, to three places.  Fine.  Except it uses
&error_abort in these places, changing the behavior from exit(1) to
abort(), and thus undoing the work of commit 39228250 "exec: Don't
abort when we can't allocate guest memory".

The three places are:

* memory_region_init_ram()

  Commit 49946538 (right after commit ef701d7b) lifted the error
  handling further, through memory_region_init_ram(), multiplying the
  incorrect use of &error_abort.  Later on, imitation of existing
  (bad) code may have created more.

* memory_region_init_ram_ptr()

  The &error_abort is still there.

* memory_region_init_rom_device()

  Doesn't need fixing, because commit 33e0eb52 (soon after commit
  ef701d7b) lifted the error handling further, and in the process
  changed it from &error_abort to passing it up the call chain.
  Correct, because the callers are realize() methods.

Fix the error handling after memory_region_init_ram() with a
Coccinelle semantic patch:

    @r@
    expression mr, owner, name, size, err;
    position p;
    @@
            memory_region_init_ram(mr, owner, name, size,
    (
    -                              &error_abort
    +                              &error_fatal
    |
                                   err@p
    )
                                  );
    @script:python@
        p << r.p;
    @@
    print "%s:%s:%s" % (p[0].file, p[0].line, p[0].column)

When the last argument is &error_abort, it gets replaced by
&error_fatal.  This is the fix.

If the last argument is anything else, its position is reported.  This
lets us check the fix is complete.  Four positions get reported:

* ram_backend_memory_alloc()

  Error is passed up the call chain, ultimately through
  user_creatable_complete().  As far as I can tell, it's callers all
  handle the error sanely.

* fsl_imx25_realize(), fsl_imx31_realize(), dp8393x_realize()

  DeviceClass.realize() methods, errors handled sanely further up the
  call chain.

We're good.  Test case again behaves:

    $ qemu-system-x86_64 -m 10000000
    qemu-system-x86_64: cannot set up guest memory 'pc.ram': Cannot allocate memory
    [Exit 1 ]

The next commits will repair the rest of commit ef701d7b's damage.
Signed-off-by: NMarkus Armbruster <armbru@redhat.com>
Message-Id: <1441983105-26376-3-git-send-email-armbru@redhat.com>
Reviewed-by: NPeter Crosthwaite <crosthwaite.peter@gmail.com>
      f8ed85ac
  17. 10 9月, 2015 4 次提交
  19. 08 9月, 2015 1 次提交
  21. 24 6月, 2015 1 次提交
    • S
      Revert "xen-hvm: increase maxmem before calling xc_domain_populate_physmap" · ffffbb36
      由 Stefano Stabellini 提交于 
      This reverts commit c1d322e6.

The original commit fixes a bug when assigning a large number of
devices which require option roms to a guest.  (One known
configuration that needs extra memory is having more than 4 emulated
NICs assigned.  Three or fewer NICs seems to work without this
functionality.)

However, by unilaterally increasing maxmem, it introduces two
problems.

First, now libxl's calculation of the required maxmem during migration
is broken -- any guest which exercised this functionality will fail on
migration.  (Guests which have the default number of devices are not
affected.)

Secondly, it makes it impossible for a higher-level toolstack or
administer to predict how much memory a VM will actually use, making
it much more difficult to effectively use all of the memory on a
machine.
Signed-off-by: NGeorge Dunlap <george.dunlap@eu.citrix.com>
Signed-off-by: NStefano Stabellini <stefano.stabellini@eu.citrix.com>
      ffffbb36
  23. 05 6月, 2015 2 次提交
    • P
      memory: prepare for multiple bits in the dirty log mask · b2dfd71c
      由 Paolo Bonzini 提交于 
      When the dirty log mask will also cover other bits than DIRTY_MEMORY_VGA,
some listeners may be interested in the overall zero/non-zero value of
the dirty log mask; others may be interested in the value of single bits.

For this reason, always call log_start/log_stop if bits have respectively
appeared or disappeared, and pass the old and new values of the dirty log
mask so that listeners can distinguish the kinds of change.

For example, KVM checks if dirty logging used to be completely disabled
(in log_start) or is now completely disabled (in log_stop).  On the
other hand, Xen has to check manually if DIRTY_MEMORY_VGA changed,
since that is the only bit it cares about.
Reviewed-by: NFam Zheng <famz@redhat.com>
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>
      b2dfd71c
    • P
      memory: differentiate memory_region_is_logging and memory_region_get_dirty_log_mask · 2d1a35be
      由 Paolo Bonzini 提交于 
      For now memory regions only track DIRTY_MEMORY_VGA individually, but
this will change soon.  To support this, split memory_region_is_logging
in two functions: one that returns a given bit from dirty_log_mask,
and one that returns the entire mask.  memory_region_is_logging gets an
extra parameter so that the compiler flags misuse.

While VGA-specific users (including the Xen listener!) will want to keep
checking that bit, KVM and vhost check for "any bit except migration"
(because migration is handled via the global start/stop listener
callbacks).
Reviewed-by: NFam Zheng <famz@redhat.com>
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>
      2d1a35be
  25. 20 1月, 2015 1 次提交
  27. 14 1月, 2015 1 次提交
  29. 11 11月, 2014 1 次提交
  31. 30 10月, 2014 1 次提交
  33. 20 9月, 2014 1 次提交
  35. 09 9月, 2014 1 次提交
  37. 28 8月, 2014 2 次提交
  39. 01 8月, 2014 1 次提交
  41. 18 7月, 2014 1 次提交
  43. 07 7月, 2014 1 次提交
  45. 23 6月, 2014 2 次提交
  47. 11 6月, 2014 1 次提交
  49. 08 5月, 2014 2 次提交
  51. 06 5月, 2014 1 次提交
  53. 05 3月, 2014 1 次提交