Greg Kurz [Thu, 1 Feb 2018 20:21:28 +0000 (21:21 +0100)]
tests: virtio-9p: add WRITE operation test
Trivial test of a successful write.
Signed-off-by: Greg Kurz <[email protected]>
(groug, handle potential overflow when computing request size,
add missing g_free(buf),
backend handles one written byte at a time to validate
the server doesn't do short-reads) Reviewed-by: Stefan Hajnoczi <[email protected]>
Greg Kurz [Thu, 1 Feb 2018 20:21:27 +0000 (21:21 +0100)]
tests: virtio-9p: use the synth backend
The purpose of virtio-9p-test is to test the virtio-9p device, especially
the 9p server state machine. We don't really care what fsdev backend we're
using. Moreover, if we want to be able to test the flush request or a
device reset with in-flights I/O, it is close to impossible to achieve
with a physical backend because we cannot ask it reliably to put an I/O
on hold at a specific point in time.
Fortunately, we can do that with the synthetic backend, which allows to
register callbacks on read/write accesses to a specific file. This will
be used by a later patch to test the 9P flush request.
The walk request test is converted to using the synth backend.
Greg Kurz [Thu, 1 Feb 2018 20:21:27 +0000 (21:21 +0100)]
tests: virtio-9p: wait for completion in the test code
In order to test request cancellation, we will need to send multiple
requests and wait for the associated replies. Since we poll the ISR
to know if a request completed, we may have several replies to parse
when we detect ISR was set to 1.
This patch moves the waiting out of the reply parsing path, up into
the functional tests.
Greg Kurz [Thu, 1 Feb 2018 20:21:27 +0000 (21:21 +0100)]
tests: virtio-9p: move request tag to the test functions
It doesn't really makes sense to hide the request tag from the test
functions. It prevents to test the 9p server behavior when passed
a wrong tag (ie, still in use or different from P9_NOTAG for a
version request). Also the spec says that a tag is reusable as soon
as the corresponding request was replied or flushed: no need to
always increment tags like we do now. And finaly, an upcoming test
of the flush command will need to manipulate tags explicitely.
This simply changes all request functions to have a tag argument.
Except for the version request which needs P9_NOTAG, all other
tests can pass 0 since they wait for the reply before sending
another request.
Keno Fischer [Thu, 1 Feb 2018 20:21:27 +0000 (21:21 +0100)]
9pfs: Correctly handle cancelled requests
# Background
I was investigating spurious non-deterministic EINTR returns from
various 9p file system operations in a Linux guest served from the
qemu 9p server.
## EINTR, ERESTARTSYS and the linux kernel
When a signal arrives that the Linux kernel needs to deliver to user-space
while a given thread is blocked (in the 9p case waiting for a reply to its
request in 9p_client_rpc -> wait_event_interruptible), it asks whatever
driver is currently running to abort its current operation (in the 9p case
causing the submission of a TFLUSH message) and return to user space.
In these situations, the error message reported is generally ERESTARTSYS.
If the userspace processes specified SA_RESTART, this means that the
system call will get restarted upon completion of the signal handler
delivery (assuming the signal handler doesn't modify the process state
in complicated ways not relevant here). If SA_RESTART is not specified,
ERESTARTSYS gets translated to EINTR and user space is expected to handle
the restart itself.
## The 9p TFLUSH command
The 9p TFLUSH commands requests that the server abort an ongoing operation.
The man page [1] specifies:
```
If it recognizes oldtag as the tag of a pending transaction, it should
abort any pending response and discard that tag.
[...]
When the client sends a Tflush, it must wait to receive the corresponding
Rflush before reusing oldtag for subsequent messages. If a response to the
flushed request is received before the Rflush, the client must honor the
response as if it had not been flushed, since the completed request may
signify a state change in the server
```
In particular, this means that the server must not send a reply with the
orignal tag in response to the cancellation request, because the client is
obligated to interpret such a reply as a coincidental reply to the original
request.
# The bug
When qemu receives a TFlush request, it sets the `cancelled` flag on the
relevant pdu. This flag is periodically checked, e.g. in
`v9fs_co_name_to_path`, and if set, the operation is aborted and the error
is set to EINTR. However, the server then violates the spec, by returning
to the client an Rerror response, rather than discarding the message
entirely. As a result, the client is required to assume that said Rerror
response is a result of the original request, not a result of the
cancellation and thus passes the EINTR error back to user space.
This is not the worst thing it could do, however as discussed above, the
correct error code would have been ERESTARTSYS, such that user space
programs with SA_RESTART set get correctly restarted upon completion of
the signal handler.
Instead, such programs get spurious EINTR results that they were not
expecting to handle.
It should be noted that there are plenty of user space programs that do not
set SA_RESTART and do not correctly handle EINTR either. However, that is
then a userspace bug. It should also be noted that this bug has been
mitigated by a recent commit to the Linux kernel [2], which essentially
prevents the kernel from sending Tflush requests unless the process is about
to die (in which case the process likely doesn't care about the response).
Nevertheless, for older kernels and to comply with the spec, I believe this
change is beneficial.
# Implementation
The fix is fairly simple, just skipping notification of a reply if
the pdu was previously cancelled. We do however, also notify the transport
layer that we're doing this, so it can clean up any resources it may be
holding. I also added a new trace event to distinguish
operations that caused an error reply from those that were cancelled.
One complication is that we only omit sending the message on EINTR errors in
order to avoid confusing the rest of the code (which may assume that a
client knows about a fid if it sucessfully passed it off to pud_complete
without checking for cancellation status). This does mean that if the server
acts upon the cancellation flag, it always needs to set err to EINTR. I
believe this is true of the current code.
Peter Maydell [Wed, 31 Jan 2018 15:50:29 +0000 (15:50 +0000)]
Merge remote-tracking branch 'remotes/rth/tags/pull-hppa-20180131' into staging
Implement hppa-softmmu
# gpg: Signature made Wed 31 Jan 2018 14:19:06 GMT
# gpg: using RSA key 0x64DF38E8AF7E215F
# gpg: Good signature from "Richard Henderson <[email protected]>"
# Primary key fingerprint: 7A48 1E78 868B 4DB6 A85A 05C0 64DF 38E8 AF7E 215F
* remotes/rth/tags/pull-hppa-20180131: (43 commits)
target/hppa: Implement PROBE for system mode
target/hppa: Fix 32-bit operand masks for 0E FCVT
hw/hppa: Add MAINTAINERS entry
pc-bios: Add hppa-firmware.img and git submodule
hw/hppa: Implement DINO system board
target/hppa: Enable MTTCG
target/hppa: Implement STWA
target/hppa: Implement a pause instruction
target/hppa: Implement LDSID for system mode
target/hppa: Fix comment
target/hppa: Increase number of temp regs
target/hppa: Only use EXCP_DTLB_MISS
target/hppa: Implement B,GATE insn
target/hppa: Add migration for the cpu
target/hppa: Add system registers to gdbstub
target/hppa: Optimize for flat addressing space
target/hppa: Implement halt and reset instructions
target/hppa: Implement SYNCDMA insn
target/hppa: Implement LCI
target/hppa: Implement LPA
...
Helge Deller [Sun, 8 Oct 2017 20:47:27 +0000 (16:47 -0400)]
hw/hppa: Implement DINO system board
Now that we have the prerequisites in target/hppa/,
implement the hardware for a PA7100LC.
This also enables build for hppa-softmmu.
Signed-off-by: Helge Deller <[email protected]>
[rth: Since it is all new code, squashed all branch development
withing hw/hppa/ to a single patch.] Signed-off-by: Richard Henderson <[email protected]>
Unknown why this works, but if we return EXCP_ITLB_MISS we
will triple-fault the first userland instruction fetch.
Is it something to do with having a combined I/DTLB?
Helge Deller [Fri, 29 Dec 2017 06:04:57 +0000 (22:04 -0800)]
target/hppa: Implement halt and reset instructions
Real hardware would use an external device to control the power.
But for the moment let's invent instructions in reserved space,
to be used by our custom firmware.
target/hppa: Split address size from register size
For system mode, we will need 64-bit virtual addresses even when
we have 32-bit register sizes. Since the rest of QEMU equates
TARGET_LONG_BITS with the address size, redefine everything
related to register size in terms of a new TARGET_REGISTER_BITS.
Peter Maydell [Tue, 30 Jan 2018 15:20:01 +0000 (15:20 +0000)]
Merge remote-tracking branch 'remotes/stefanberger/tags/pull-tpm-2018-01-26-2' into staging
Merge tpm 2018/01/26 v2
# gpg: Signature made Mon 29 Jan 2018 22:20:05 GMT
# gpg: using RSA key 0x75AD65802A0B4211
# gpg: Good signature from "Stefan Berger <[email protected]>"
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg: There is no indication that the signature belongs to the owner.
# Primary key fingerprint: B818 B9CA DF90 89C2 D5CE C66B 75AD 6580 2A0B 4211
* remotes/stefanberger/tags/pull-tpm-2018-01-26-2:
tpm: add CRB device
tpm: report backend request error
tpm: replace GThreadPool with AIO threadpool
tpm: lookup cancel path under tpm device class
tpm: fix alignment issues
tpm: Set the flags of the CMD_INIT command to 0
Peter Maydell [Tue, 30 Jan 2018 13:17:19 +0000 (13:17 +0000)]
linux-user/signal.c: Rename MC_* defines
The SPARC code in linux-user/signal.c defines a set of
MC_* constants. On some SPARC hosts these are also defined
by sys/ucontext.h, resulting in build failures:
In file included from /usr/include/signal.h:302:0,
from include/qemu/osdep.h:86,
from linux-user/signal.c:19:
/usr/include/sparc64-linux-gnu/sys/ucontext.h:59:0: note: this is the location of the previous definition
# define MC_NGREG __MC_NGREG
Rename all these constants to SPARC_MC_* to avoid the clash.
Peter Maydell [Tue, 30 Jan 2018 09:47:51 +0000 (09:47 +0000)]
Merge remote-tracking branch 'remotes/stefanha/tags/tracing-pull-request' into staging
# gpg: Signature made Mon 29 Jan 2018 15:49:05 GMT
# gpg: using RSA key 0x9CA4ABB381AB73C8
# gpg: Good signature from "Stefan Hajnoczi <[email protected]>"
# gpg: aka "Stefan Hajnoczi <[email protected]>"
# Primary key fingerprint: 8695 A8BF D3F9 7CDA AC35 775A 9CA4 ABB3 81AB 73C8
* remotes/stefanha/tags/tracing-pull-request:
tracetool: report error on foo() instead of foo(void)
tracetool: clarify that "formats" means "format strings"
tracetool: prefix parse errors with line numbers
tpm_crb is a device for TPM 2.0 Command Response Buffer (CRB)
Interface as defined in TCG PC Client Platform TPM Profile (PTP)
Specification Family “2.0” Level 00 Revision 01.03 v22.
The PTP allows device implementation to switch between TIS and CRB
model at run time, but given that CRB is a simpler device to
implement, I chose to implement it as a different device.
The device doesn't implement other locality than 0 for now (my laptop
TPM doesn't either, so I assume this isn't so bad)
Tested with some success with Linux upstream and Windows 10, seabios &
modified ovmf. The device is recognized and correctly transmit
command/response with passthrough & emu. However, we are missing PPI
ACPI part atm.
The TPM backend uses a GThreadPool to handle IO in a seperate
thread. However, GThreadPool isn't integrated with Qemu main loops,
making it unnecessarily complicated to deal with.
Qemu has a AIO threadpool, that is better integrated with loops and
various IO functions, provides completion BH by default etc.
Remove the only user of GThreadPool from qemu, use AIO threadpool.
Note that the backend:
- no longer accepts queing multiple requests (unneeded so far)
- increase ref to itself when handling a command, for extra safety
- tpm_backend_thread_end() is renamed tpm_backend_finish_sync() and
will wait for completion of BH (request_completed), which will help
migration handling.
Since Linux commit 313d21eeab9282e, tpm devices have their own device
class "tpm" and the cancel path must be looked up under
/sys/class/tpm/ instead of /sys/class/misc/.
TEST: tests/tpm-crb-test... (pid=230409)
/i386/tpm-crb/test:
Broken pipe
FAIL
GTester: last random seed: R02S29cea50247fe1efa59ee885a26d51a85
(pid=230423)
FAIL: tests/tpm-crb-test
and generates a new clang sanitizer runtime warning:
/home/petmay01/linaro/qemu-for-merges/hw/tpm/tpm_util.h:36:24: runtime
error: load of misaligned address 0x7fdc24c00002 for type 'const
uint32_t' (aka 'const unsigned int'), which requires 4 byte alignment
0x7fdc24c00002: note: pointer points here
<memory cannot be printed>
The sparc architecture does not allow misaligned loads and will
segfault if you try them. For example, this function:
As a general rule you can't take an arbitrary pointer into a byte
buffer and try to interpret it as a structure or a pointer to a
larger-than-bytesize-data simply by casting the pointer.
Use this clean up as an opportunity to remove unnecessary temporary
buffers and casts.
Peter Maydell [Mon, 29 Jan 2018 14:29:17 +0000 (14:29 +0000)]
Merge remote-tracking branch 'remotes/jasowang/tags/net-pull-request' into staging
# gpg: Signature made Mon 29 Jan 2018 08:14:19 GMT
# gpg: using RSA key 0xEF04965B398D6211
# gpg: Good signature from "Jason Wang (Jason Wang on RedHat) <[email protected]>"
# gpg: WARNING: This key is not certified with sufficiently trusted signatures!
# gpg: It is not certain that the signature belongs to the owner.
# Primary key fingerprint: 215D 46F4 8246 689E C77F 3562 EF04 965B 398D 6211
* remotes/jasowang/tags/net-pull-request:
MAINTAINERS: update Dmitry Fleytman email
qemu-doc: Get rid of "vlan=X" example in the documentation
net: Allow netdevs to be used with 'hostfwd_add' and 'hostfwd_remove'
net: Allow hubports to connect to other netdevs
colo: compare the packet based on the tcp sequence number
colo: modified the payload compare function
Peter Maydell [Mon, 29 Jan 2018 13:18:14 +0000 (13:18 +0000)]
Merge remote-tracking branch 'remotes/dgibson/tags/ppc-for-2.12-20180129' into staging
ppc patch queue 2018-01-29
Here's another batch of patches for ppc, spapr and related things.
Higlights:
* Implement (with a bunch of necessary infrastructure) a hypercall
to let guests properly apply Spectre and Meltdown workarounds.
* Convert a number of old devices to trace events
* Fix some bugs
* remotes/dgibson/tags/ppc-for-2.12-20180129:
target/ppc/spapr: Add H-Call H_GET_CPU_CHARACTERISTICS
target/ppc/spapr_caps: Add new tristate cap safe_indirect_branch
target/ppc/spapr_caps: Add new tristate cap safe_bounds_check
target/ppc/spapr_caps: Add new tristate cap safe_cache
target/ppc/spapr_caps: Add support for tristate spapr_capabilities
target/ppc/kvm: Add cap_ppc_safe_[cache/bounds_check/indirect_branch]
spapr_pci: fix MSI/MSIX selection
input: add missing newline from trace-events
uninorth: convert to trace-events
grackle: convert to trace-events
ppc: Deprecate qemu-system-ppcemb
ppc/pnv: fix PnvChip redefinition in <hw/ppc/pnv_xscom.h>
Stefan Hajnoczi [Wed, 10 Jan 2018 20:25:53 +0000 (20:25 +0000)]
tracetool: report error on foo() instead of foo(void)
C functions with no arguments must be declared foo(void) instead of
foo(). The tracetool argument list parser has never accepted an empty
argument list. This patch adds a clear error message for this error
case.
Stefan Hajnoczi [Wed, 10 Jan 2018 20:25:52 +0000 (20:25 +0000)]
tracetool: clarify that "formats" means "format strings"
The terminology used by tracetool is not consistent with C sprintf or
docs/devel/tracing.txt. The word "formats" is sometimes used to mean
"format strings".
This patch clarifies comments and error messages that contain this word.
Note that the error message lines are longer than 80 characters but I
have not wrapped them to aid grepping.
NB, the virtio-input device reports a bitmask to the guest driver that
has a bit set for each Linux keycode that the host is able to send to
the guest.
Thus by adding these extra key mappings we are technically changing the
host<->guest ABI. This would also happen any time we defined new mappings
for QEMU keycodes in future.
When a keycode is removed from the list of possible keycodes that host can
send to the guest, it means that the guest OS will think it is possible
to receive a key that in pratice can never be generated, which is harmless.
When a keycode is added to the list of possible keycodes that the host can
send to the guest, it means that the guest OS can see an unexpected event.
The Linux virtio_input.c driver code simply forwards this event to the
input_event() method in the Linux input subsystem. This in turn calls
input_handle_event(), which then calls input_get_disposition(). This method
checks if the input event is present in the permitted keys bitmap, and if
not returns INPUT_IGNORE_EVENT. Thus the unexpected event will get dropped,
which is harmless.
If the guest OS reboots, or otherwise re-initializes the virt-input device,
it will read the new keycode bitmap. No matter how many keys are defined,
the config space has a fixed 128 byte bitmap. There is, however, a size
field defiend which says how many bytes in the bitmap are used. So the guest
OS reads the size of the bitmap, and then it reads the data from bitmap upto
the designated size. So if the guest OS re-initializes at precisely the time
that QEMU is migrated across versions, in the worst case, it could conceivably
read the old size field, but then get the newly updated bitmap. If a key were
added this is harmless, since it simply means it may not process the newly
added key. If a key were removed, then it could be readnig a byte from the
bitmap that was not initialized. Fortunately QEMU always memsets() the entire
bitmap to 0, prior to setting keybits. Thus the guest OS will simply read
zeros, which is again harmless.
Based on this analysis, it is believed that there is no need to preserve the
virtio-input-hid keymaps across migration, as the host<->guest ABI change is
harmless and self-resolving at time of guest reboot.
NB, this behaviour should perhaps be formalized in the virtio-input spec
to declare how guest OS drivers should be written to be robust in their
handling of the potentially changable key bitmaps.
- Q_KEY_CODE_HIRAGANA was mapped to 0x70 (Katakanahiragana)
instead of of 0x77 (Hirigana)
- Q_KEY_CODE_MENU was incorrectly mapped to the compose
scancode (0xe05d) and is now mapped to 0xe01e
- Q_KEY_CODE_FIND was mapped to 0xe065 (Search) instead
of to 0xe041 (Find)
- Q_KEY_CODE_POWER, SLEEP & WAKE had 0x0e instead of 0xe0
as the prefix
Missing entries in qcode_to_keycode_set2 now fixed:
- Q_KEY_CODE_PRINT -> 0x7f (NB ignored due to special case)
- Q_KEY_CODE_COMPOSE -> 0xe02f
- Q_KEY_CODE_PAUSE -> 0xe077
- Q_KEY_CODE_KP_EQUALS -> 0x0f
And some mistakes corrected:
- Q_KEY_CODE_HIRAGANA was mapped to 0x13 (Katakanahiragana)
instead of of 0x62 (Hirigana)
- Q_KEY_CODE_MENU was incorrectly mapped to the compose
scancode (0xe02f) and is now not mapped
- Q_KEY_CODE_FIND was mapped to 0xe010 (Search) and is now
not mapped.
- Q_KEY_CODE_POWER, SLEEP & WAKE had 0x0e instead of 0xe0
as the prefix
Missing entries in qcode_to_keycode_set3 now fixed:
During Qemu guest migration, a destination process invokes ps2
post_load function. In that, if 'rptr' and 'count' values were
invalid, it could lead to OOB access or infinite loop issue.
Add check to avoid it.
Miika S [Fri, 22 Dec 2017 15:25:31 +0000 (17:25 +0200)]
input: virtio: don't send mouse wheel event twice
On Linux, a mouse event is generated for both down and up when mouse
wheel is used. This caused virtio_input_send() to be called twice each
time the wheel was used.
This commit adds a check for the button down state and only calls
virtio_input_send() when it is true.
Thomas Huth [Thu, 11 Jan 2018 20:02:40 +0000 (21:02 +0100)]
net: Allow netdevs to be used with 'hostfwd_add' and 'hostfwd_remove'
It does not make much sense to limit these commands to the legacy 'vlan'
concept only, they should work with the modern netdevs, too. So now
it is possible to use this command with one, two or three parameters.
With one parameter, the command installs a hostfwd rule on the default
"user" network:
hostfwd_add tcp:...
With two parameters, the command installs a hostfwd rule on a netdev
(that's the new way of using this command):
hostfwd_add netdev_id tcp:...
With three parameters, the command installs a rule on a 'vlan' (aka hub):
hostfwd_add hub_id name tcp:...
Thomas Huth [Mon, 15 Jan 2018 19:50:55 +0000 (20:50 +0100)]
net: Allow hubports to connect to other netdevs
QEMU can emulate hubs to connect NICs and netdevs. This is currently
primarily used for the mis-named 'vlan' feature of the networking
subsystem. Now the 'vlan' feature has been marked as deprecated, since
its name is rather confusing and the users often rather mis-configure
their network when trying to use it. But while the 'vlan' parameter
should be removed at one point in time, the basic idea of emulating
a hub in QEMU is still good: It's useful for bundling up the output of
multiple NICs into one single l2tp netdev for example.
Now to be able to use the hubport feature without 'vlan's, there is one
missing piece: The possibility to connect a hubport to a netdev, too.
This patch adds this possibility by introducing a new "netdev=..."
parameter to the hubports.
To bundle up the output of multiple NICs into one socket netdev, you can
now run QEMU with these parameters for example:
After the ppc64 guest tried to boot from both NICs, you can see in the
dump file (using Wireshark, for example), that the output of both NICs
(the e1000 and the virtio-net-pci) has been successfully transfered
via the socket netdev in this case.
Mao Zhongyi [Mon, 25 Dec 2017 02:54:12 +0000 (10:54 +0800)]
colo: compare the packet based on the tcp sequence number
Packet size some time different or when network is busy.
Based on same payload size, but TCP protocol can not
guarantee send the same one packet in the same way,
like that:
We send this payload:
------------------------------
| header |1|2|3|4|5|6|7|8|9|0|
------------------------------
In the original method, ppkt1 and ppkt2 are different in size and
spkt1, so they can't compare and trigger the checkpoint.
I have tested FTP get 200M and 1G file many times, I found that
the performance was less than 1% of the native.
Now I reconstructed the comparison of TCP packets based on the
TCP sequence number. first of all, ppkt1 and spkt1 have the same
starting sequence number, so they can compare, even though their
length is different. And then ppkt1 with a smaller payload length
is used as the comparison length, if the payload is same, send
out the ppkt1 and record the offset(the length of ppkt1 payload)
in spkt1. The next comparison, ppkt2 and spkt1 can be compared
from the recorded position of spkt1.
target/ppc/spapr_caps: Add support for tristate spapr_capabilities
spapr_caps are used to represent the level of support for various
capabilities related to the spapr machine type. Currently there is
only support for boolean capabilities.
Add support for tristate capabilities by implementing their get/set
functions. These capabilities can have the values 0, 1 or 2
corresponding to broken, workaround and fixed.
Add three new kvm capabilities used to represent the level of host support
for three corresponding workarounds.
Host support for each of the capabilities is queried through the
new ioctl KVM_PPC_GET_CPU_CHAR which returns four uint64 quantities. The
first two, character and behaviour, represent the available
characteristics of the cpu and the behaviour of the cpu respectively.
The second two, c_mask and b_mask, represent the mask of known bits for
the character and beheviour dwords respectively.
Signed-off-by: Suraj Jitindar Singh <[email protected]> Reviewed-by: David Gibson <[email protected]>
[dwg: Correct some compile errors due to name change in final kernel
patch version] Signed-off-by: David Gibson <[email protected]>
Greg Kurz [Fri, 26 Jan 2018 22:25:24 +0000 (23:25 +0100)]
spapr_pci: fix MSI/MSIX selection
In various place we don't correctly check if the device supports MSI or
MSI-X. This can cause devices to be advertised with MSI support, even
if they only support MSI-X (like virtio-pci-* devices for example):
Worse, this can also cause the "ibm,change-msi" RTAS call to corrupt the
PCI status and cause migration to fail:
qemu-system-ppc64: get_pci_config_device: Bad config data: i=0x6
read: 0 device: 10 cmask: 10 wmask: 0 w1cmask:0
^^
PCI_STATUS_CAP_LIST bit which is assumed to be constant
This patch changes spapr_populate_pci_child_dt() to properly check for
MSI support using msi_present(): this ensures that PCIDevice::msi_cap
was set by msi_init() and that msi_nr_vectors_allocated() will look at
the right place in the config space.
Checking PCIDevice::msix_entries_nr is enough for MSI-X but let's add
a call to msix_present() there as well for consistency.
It also changes rtas_ibm_change_msi() to select the appropriate MSI
type in Function 1 instead of always selecting plain MSI. This new
behaviour is compliant with LoPAPR 1.1, as described in "Table 71.
ibm,change-msi Argument Call Buffer":
Function 1: If Number Outputs is equal to 3, request to set to a new
number of MSIs (including set to 0).
If the “ibm,change-msix-capable” property exists and Number
Outputs is equal to 4, request is to set to a new number of
MSI or MSI-X (platform choice) interrupts (including set to
0).
Since MSI is the the platform default (LoPAPR 6.2.3 MSI Option), let's
check for MSI support first.
And finally, it checks the input parameters are valid, as described in
LoPAPR 1.1 "R1–7.3.10.5.1–3":
For the MSI option: The platform must return a Status of -3 (Parameter
error) from ibm,change-msi, with no change in interrupt assignments if
the PCI configuration address does not support MSI and Function 3 was
requested (that is, the “ibm,req#msi” property must exist for the PCI
configuration address in order to use Function 3), or does not support
MSI-X and Function 4 is requested (that is, the “ibm,req#msi-x” property
must exist for the PCI configuration address in order to use Function 4),
or if neither MSIs nor MSI-Xs are supported and Function 1 is requested.
This ensures that the ret_intr_type variable contains a valid MSI type
for this device, and that spapr_msi_setmsg() won't corrupt the PCI status.
qemu-system-ppcemb has been once split of qemu-system-ppc to support
CPU page sizes < 4096 for some of the embedded 4xx PowerPC CPUs.
However, there was hardly any OS available in the wild that really
used such small page sizes (Linux uses 4096 on PPC), so there is
no known recent use case for this separate build anymore. It's
rather cumbersome to maintain a separate set of config switches for
this, and it's wasting compile and test time of all the developers
who have to build all QEMU targets to verify that their changes did
not break anything.
Except for the small CPU page sizes, qemu-system-ppc can be used as
a full replacement for qemu-system-ppcemb since it contains all the
embedded 4xx PPC boards and CPUs, too. Thus let's start the deprecation
process for qemu-system-ppcemb to see whether somebody still needs
the small page sizes or whether we could finally remove this unloved
separate build.
Cédric Le Goater [Fri, 19 Jan 2018 08:15:11 +0000 (09:15 +0100)]
ppc/pnv: fix PnvChip redefinition in <hw/ppc/pnv_xscom.h>
This redefinition generates warnings on some clang compilers and older
gcc4.4.
...include/hw/ppc/pnv_xscom.h:24:24: warning: redefinition of typedef 'PnvChip' is a C11
feature [-Wtypedef-redefinition]
typedef struct PnvChip PnvChip;
^
...include/hw/ppc/pnv.h:65:3: note: previous definition is here
} PnvChip;
^
1 warning generated.
CC ppc64-softmmu/hw/ppc/pnv_xscom.o
Peter Maydell [Fri, 26 Jan 2018 17:29:14 +0000 (17:29 +0000)]
Merge remote-tracking branch 'remotes/ericb/tags/pull-nbd-2018-01-26' into staging
nbd patches for 2018-01-26
- Vladimir Sementsov-Ogievskiy - nbd export qmp interface
- Eric Blake - hmp: Add nbd_server_remove to mirror QMP command
- Edgar Kaziakhmedov - nbd: implement bdrv_get_info callback
# gpg: Signature made Fri 26 Jan 2018 16:02:34 GMT
# gpg: using RSA key 0xA7A16B4A2527436A
# gpg: Good signature from "Eric Blake <[email protected]>"
# gpg: aka "Eric Blake (Free Software Programmer) <[email protected]>"
# gpg: aka "[jpeg image of size 6874]"
# Primary key fingerprint: 71C2 CC22 B1C4 6029 27D2 F3AA A7A1 6B4A 2527 436A
* remotes/ericb/tags/pull-nbd-2018-01-26:
nbd: implement bdrv_get_info callback
hmp: Add nbd_server_remove to mirror QMP command
iotest 205: new test for qmp nbd-server-remove
iotests: implement QemuIoInteractive class
iotest 147: add cases to test new @name parameter of nbd-server-add
qapi: add nbd-server-remove
hmp: Add name parameter to nbd_server_add
qapi: add name parameter to nbd-server-add
Since mirror job supports efficient zero out target mechanism (see
in mirror_dirty_init()), implement bdrv_get_info to make it work
over NBD. Such improvement will allow using the largest chunk possible
and will decrease the number of NBD_CMD_WRITE_ZEROES requests on the wire.
Eric Blake [Thu, 25 Jan 2018 14:45:57 +0000 (08:45 -0600)]
hmp: Add nbd_server_remove to mirror QMP command
Since everything else about the nbd-server-* QMP commands is
accessible from HMP, we might as well make removing an export
available as well. For now, I went with a bool flag rather
than a mode string for choosing between safe (default) and
hard modes.
Add command for removing an export. It is needed for cases when we
don't want to keep the export after the operation on it was completed.
The other example is a temporary node, created with blockdev-add.
If we want to delete it we should firstly remove any corresponding
NBD export.