* [Qemu-devel] [PULL 01/10] 9pfs: drop v9fs_register_transport()
2018-01-30 17:39 [Qemu-devel] [PULL 00/10] 9p patches for 2.12 20180130 Greg Kurz
@ 2018-01-30 17:39 ` Greg Kurz
2018-01-30 17:39 ` [Qemu-devel] [PULL 02/10] 9pfs: Correctly handle cancelled requests Greg Kurz
` (9 subsequent siblings)
10 siblings, 0 replies; 13+ messages in thread
From: Greg Kurz @ 2018-01-30 17:39 UTC (permalink / raw)
To: qemu-devel; +Cc: Peter Maydell, Greg Kurz
No good reasons to do this outside of v9fs_device_realize_common().
Signed-off-by: Greg Kurz <groug@kaod.org>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>
---
hw/9pfs/9p.c | 6 +++++-
hw/9pfs/9p.h | 10 ++--------
hw/9pfs/virtio-9p-device.c | 8 ++------
hw/9pfs/xen-9p-backend.c | 3 +--
4 files changed, 10 insertions(+), 17 deletions(-)
diff --git a/hw/9pfs/9p.c b/hw/9pfs/9p.c
index 909a61139405..364c7cb44628 100644
--- a/hw/9pfs/9p.c
+++ b/hw/9pfs/9p.c
@@ -3485,7 +3485,8 @@ void pdu_submit(V9fsPDU *pdu, P9MsgHeader *hdr)
}
/* Returns 0 on success, 1 on failure. */
-int v9fs_device_realize_common(V9fsState *s, Error **errp)
+int v9fs_device_realize_common(V9fsState *s, const V9fsTransport *t,
+ Error **errp)
{
int i, len;
struct stat stat;
@@ -3493,6 +3494,9 @@ int v9fs_device_realize_common(V9fsState *s, Error **errp)
V9fsPath path;
int rc = 1;
+ assert(!s->transport);
+ s->transport = t;
+
/* initialize pdu allocator */
QLIST_INIT(&s->free_list);
QLIST_INIT(&s->active_list);
diff --git a/hw/9pfs/9p.h b/hw/9pfs/9p.h
index ffe658ab8975..5ced427d861b 100644
--- a/hw/9pfs/9p.h
+++ b/hw/9pfs/9p.h
@@ -346,7 +346,8 @@ void v9fs_path_sprintf(V9fsPath *path, const char *fmt, ...);
void v9fs_path_copy(V9fsPath *lhs, V9fsPath *rhs);
int v9fs_name_to_path(V9fsState *s, V9fsPath *dirpath,
const char *name, V9fsPath *path);
-int v9fs_device_realize_common(V9fsState *s, Error **errp);
+int v9fs_device_realize_common(V9fsState *s, const V9fsTransport *t,
+ Error **errp);
void v9fs_device_unrealize_common(V9fsState *s, Error **errp);
V9fsPDU *pdu_alloc(V9fsState *s);
@@ -366,11 +367,4 @@ struct V9fsTransport {
void (*push_and_notify)(V9fsPDU *pdu);
};
-static inline int v9fs_register_transport(V9fsState *s, const V9fsTransport *t)
-{
- assert(!s->transport);
- s->transport = t;
- return 0;
-}
-
#endif
diff --git a/hw/9pfs/virtio-9p-device.c b/hw/9pfs/virtio-9p-device.c
index 43f4e53f336f..775e8ff76671 100644
--- a/hw/9pfs/virtio-9p-device.c
+++ b/hw/9pfs/virtio-9p-device.c
@@ -198,17 +198,13 @@ static void virtio_9p_device_realize(DeviceState *dev, Error **errp)
V9fsVirtioState *v = VIRTIO_9P(dev);
V9fsState *s = &v->state;
- if (v9fs_device_realize_common(s, errp)) {
- goto out;
+ if (v9fs_device_realize_common(s, &virtio_9p_transport, errp)) {
+ return;
}
v->config_size = sizeof(struct virtio_9p_config) + strlen(s->fsconf.tag);
virtio_init(vdev, "virtio-9p", VIRTIO_ID_9P, v->config_size);
v->vq = virtio_add_queue(vdev, MAX_REQ, handle_9p_output);
- v9fs_register_transport(s, &virtio_9p_transport);
-
-out:
- return;
}
static void virtio_9p_device_unrealize(DeviceState *dev, Error **errp)
diff --git a/hw/9pfs/xen-9p-backend.c b/hw/9pfs/xen-9p-backend.c
index df2a4100bf55..14f0d6a50e75 100644
--- a/hw/9pfs/xen-9p-backend.c
+++ b/hw/9pfs/xen-9p-backend.c
@@ -446,7 +446,6 @@ static int xen_9pfs_connect(struct XenDevice *xendev)
xen_9pdev->id = s->fsconf.fsdev_id =
g_strdup_printf("xen9p%d", xendev->dev);
xen_9pdev->tag = s->fsconf.tag = xenstore_read_fe_str(xendev, "tag");
- v9fs_register_transport(s, &xen_9p_transport);
fsdev = qemu_opts_create(qemu_find_opts("fsdev"),
s->fsconf.tag,
1, NULL);
@@ -455,7 +454,7 @@ static int xen_9pfs_connect(struct XenDevice *xendev)
qemu_opt_set(fsdev, "security_model", xen_9pdev->security_model, NULL);
qemu_opts_set_id(fsdev, s->fsconf.fsdev_id);
qemu_fsdev_add(fsdev);
- v9fs_device_realize_common(s, NULL);
+ v9fs_device_realize_common(s, &xen_9p_transport, NULL);
return 0;
--
2.13.6
^ permalink raw reply related [flat|nested] 13+ messages in thread* [Qemu-devel] [PULL 02/10] 9pfs: Correctly handle cancelled requests
2018-01-30 17:39 [Qemu-devel] [PULL 00/10] 9p patches for 2.12 20180130 Greg Kurz
2018-01-30 17:39 ` [Qemu-devel] [PULL 01/10] 9pfs: drop v9fs_register_transport() Greg Kurz
@ 2018-01-30 17:39 ` Greg Kurz
2018-01-30 17:39 ` [Qemu-devel] [PULL 03/10] tests: virtio-9p: move request tag to the test functions Greg Kurz
` (8 subsequent siblings)
10 siblings, 0 replies; 13+ messages in thread
From: Greg Kurz @ 2018-01-30 17:39 UTC (permalink / raw)
To: qemu-devel; +Cc: Peter Maydell, Greg Kurz, Keno Fischer
From: Keno Fischer <keno@juliacomputing.com>
# 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.
[1] https://9fans.github.io/plan9port/man/man9/flush.html
[2] https://github.com/torvalds/linux/commit/9523feac272ccad2ad8186ba4fcc891
Signed-off-by: Keno Fischer <keno@juliacomputing.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
[groug, send a zero-sized reply instead of detaching the buffer]
Signed-off-by: Greg Kurz <groug@kaod.org>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>
---
hw/9pfs/9p.c | 18 ++++++++++++++++++
hw/9pfs/trace-events | 1 +
2 files changed, 19 insertions(+)
diff --git a/hw/9pfs/9p.c b/hw/9pfs/9p.c
index 364c7cb44628..e88bb50f1365 100644
--- a/hw/9pfs/9p.c
+++ b/hw/9pfs/9p.c
@@ -630,6 +630,24 @@ static void coroutine_fn pdu_complete(V9fsPDU *pdu, ssize_t len)
V9fsState *s = pdu->s;
int ret;
+ /*
+ * The 9p spec requires that successfully cancelled pdus receive no reply.
+ * Sending a reply would confuse clients because they would
+ * assume that any EINTR is the actual result of the operation,
+ * rather than a consequence of the cancellation. However, if
+ * the operation completed (succesfully or with an error other
+ * than caused be cancellation), we do send out that reply, both
+ * for efficiency and to avoid confusing the rest of the state machine
+ * that assumes passing a non-error here will mean a successful
+ * transmission of the reply.
+ */
+ bool discard = pdu->cancelled && len == -EINTR;
+ if (discard) {
+ trace_v9fs_rcancel(pdu->tag, pdu->id);
+ pdu->size = 0;
+ goto out_notify;
+ }
+
if (len < 0) {
int err = -len;
len = 7;
diff --git a/hw/9pfs/trace-events b/hw/9pfs/trace-events
index 08a4abf22ea4..1aee350c42f1 100644
--- a/hw/9pfs/trace-events
+++ b/hw/9pfs/trace-events
@@ -1,6 +1,7 @@
# See docs/devel/tracing.txt for syntax documentation.
# hw/9pfs/virtio-9p.c
+v9fs_rcancel(uint16_t tag, uint8_t id) "tag %d id %d"
v9fs_rerror(uint16_t tag, uint8_t id, int err) "tag %d id %d err %d"
v9fs_version(uint16_t tag, uint8_t id, int32_t msize, char* version) "tag %d id %d msize %d version %s"
v9fs_version_return(uint16_t tag, uint8_t id, int32_t msize, char* version) "tag %d id %d msize %d version %s"
--
2.13.6
^ permalink raw reply related [flat|nested] 13+ messages in thread* [Qemu-devel] [PULL 03/10] tests: virtio-9p: move request tag to the test functions
2018-01-30 17:39 [Qemu-devel] [PULL 00/10] 9p patches for 2.12 20180130 Greg Kurz
2018-01-30 17:39 ` [Qemu-devel] [PULL 01/10] 9pfs: drop v9fs_register_transport() Greg Kurz
2018-01-30 17:39 ` [Qemu-devel] [PULL 02/10] 9pfs: Correctly handle cancelled requests Greg Kurz
@ 2018-01-30 17:39 ` Greg Kurz
2018-01-30 17:39 ` [Qemu-devel] [PULL 04/10] tests: virtio-9p: wait for completion in the test code Greg Kurz
` (7 subsequent siblings)
10 siblings, 0 replies; 13+ messages in thread
From: Greg Kurz @ 2018-01-30 17:39 UTC (permalink / raw)
To: qemu-devel; +Cc: Peter Maydell, Greg Kurz
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.
Signed-off-by: Greg Kurz <groug@kaod.org>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
---
tests/virtio-9p-test.c | 28 ++++++++++++++--------------
1 file changed, 14 insertions(+), 14 deletions(-)
diff --git a/tests/virtio-9p-test.c b/tests/virtio-9p-test.c
index 00f00f7246e9..5ada2839b9ae 100644
--- a/tests/virtio-9p-test.c
+++ b/tests/virtio-9p-test.c
@@ -27,7 +27,6 @@ typedef struct {
QOSState *qs;
QVirtQueue *vq;
char *test_share;
- uint16_t p9_req_tag;
} QVirtIO9P;
static QVirtIO9P *qvirtio_9p_start(const char *driver)
@@ -294,10 +293,11 @@ static void v9fs_rlerror(P9Req *req, uint32_t *err)
}
/* size[4] Tversion tag[2] msize[4] version[s] */
-static P9Req *v9fs_tversion(QVirtIO9P *v9p, uint32_t msize, const char *version)
+static P9Req *v9fs_tversion(QVirtIO9P *v9p, uint32_t msize, const char *version,
+ uint16_t tag)
{
P9Req *req = v9fs_req_init(v9p, 4 + v9fs_string_size(version), P9_TVERSION,
- P9_NOTAG);
+ tag);
v9fs_uint32_write(req, msize);
v9fs_string_write(req, version);
@@ -323,12 +323,12 @@ static void v9fs_rversion(P9Req *req, uint16_t *len, char **version)
}
/* size[4] Tattach tag[2] fid[4] afid[4] uname[s] aname[s] n_uname[4] */
-static P9Req *v9fs_tattach(QVirtIO9P *v9p, uint32_t fid, uint32_t n_uname)
+static P9Req *v9fs_tattach(QVirtIO9P *v9p, uint32_t fid, uint32_t n_uname,
+ uint16_t tag)
{
const char *uname = ""; /* ignored by QEMU */
const char *aname = ""; /* ignored by QEMU */
- P9Req *req = v9fs_req_init(v9p, 4 + 4 + 2 + 2 + 4, P9_TATTACH,
- ++(v9p->p9_req_tag));
+ P9Req *req = v9fs_req_init(v9p, 4 + 4 + 2 + 2 + 4, P9_TATTACH, tag);
v9fs_uint32_write(req, fid);
v9fs_uint32_write(req, P9_NOFID);
@@ -353,7 +353,7 @@ static void v9fs_rattach(P9Req *req, v9fs_qid *qid)
/* size[4] Twalk tag[2] fid[4] newfid[4] nwname[2] nwname*(wname[s]) */
static P9Req *v9fs_twalk(QVirtIO9P *v9p, uint32_t fid, uint32_t newfid,
- uint16_t nwname, char *const wnames[])
+ uint16_t nwname, char *const wnames[], uint16_t tag)
{
P9Req *req;
int i;
@@ -362,7 +362,7 @@ static P9Req *v9fs_twalk(QVirtIO9P *v9p, uint32_t fid, uint32_t newfid,
for (i = 0; i < nwname; i++) {
size += v9fs_string_size(wnames[i]);
}
- req = v9fs_req_init(v9p, size, P9_TWALK, ++(v9p->p9_req_tag));
+ req = v9fs_req_init(v9p, size, P9_TWALK, tag);
v9fs_uint32_write(req, fid);
v9fs_uint32_write(req, newfid);
v9fs_uint16_write(req, nwname);
@@ -397,7 +397,7 @@ static void fs_version(QVirtIO9P *v9p)
char *server_version;
P9Req *req;
- req = v9fs_tversion(v9p, P9_MAX_SIZE, version);
+ req = v9fs_tversion(v9p, P9_MAX_SIZE, version, P9_NOTAG);
v9fs_rversion(req, &server_len, &server_version);
g_assert_cmpmem(server_version, server_len, version, strlen(version));
@@ -410,7 +410,7 @@ static void fs_attach(QVirtIO9P *v9p)
P9Req *req;
fs_version(v9p);
- req = v9fs_tattach(v9p, 0, getuid());
+ req = v9fs_tattach(v9p, 0, getuid(), 0);
v9fs_rattach(req, NULL);
}
@@ -430,7 +430,7 @@ static void fs_walk(QVirtIO9P *v9p)
}
fs_attach(v9p);
- req = v9fs_twalk(v9p, 0, 1, P9_MAXWELEM, wnames);
+ req = v9fs_twalk(v9p, 0, 1, P9_MAXWELEM, wnames, 0);
v9fs_rwalk(req, &nwqid, &wqid);
g_assert_cmpint(nwqid, ==, P9_MAXWELEM);
@@ -451,7 +451,7 @@ static void fs_walk_no_slash(QVirtIO9P *v9p)
uint32_t err;
fs_attach(v9p);
- req = v9fs_twalk(v9p, 0, 1, 1, wnames);
+ req = v9fs_twalk(v9p, 0, 1, 1, wnames, 0);
v9fs_rlerror(req, &err);
g_assert_cmpint(err, ==, ENOENT);
@@ -466,10 +466,10 @@ static void fs_walk_dotdot(QVirtIO9P *v9p)
P9Req *req;
fs_version(v9p);
- req = v9fs_tattach(v9p, 0, getuid());
+ req = v9fs_tattach(v9p, 0, getuid(), 0);
v9fs_rattach(req, &root_qid);
- req = v9fs_twalk(v9p, 0, 1, 1, wnames);
+ req = v9fs_twalk(v9p, 0, 1, 1, wnames, 0);
v9fs_rwalk(req, NULL, &wqid); /* We now we'll get one qid */
g_assert_cmpmem(&root_qid, 13, wqid[0], 13);
--
2.13.6
^ permalink raw reply related [flat|nested] 13+ messages in thread* [Qemu-devel] [PULL 04/10] tests: virtio-9p: wait for completion in the test code
2018-01-30 17:39 [Qemu-devel] [PULL 00/10] 9p patches for 2.12 20180130 Greg Kurz
` (2 preceding siblings ...)
2018-01-30 17:39 ` [Qemu-devel] [PULL 03/10] tests: virtio-9p: move request tag to the test functions Greg Kurz
@ 2018-01-30 17:39 ` Greg Kurz
2018-01-30 17:39 ` [Qemu-devel] [PULL 05/10] tests: virtio-9p: use the synth backend Greg Kurz
` (6 subsequent siblings)
10 siblings, 0 replies; 13+ messages in thread
From: Greg Kurz @ 2018-01-30 17:39 UTC (permalink / raw)
To: qemu-devel; +Cc: Peter Maydell, Greg Kurz
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.
Signed-off-by: Greg Kurz <groug@kaod.org>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
---
tests/virtio-9p-test.c | 14 ++++++++++++--
1 file changed, 12 insertions(+), 2 deletions(-)
diff --git a/tests/virtio-9p-test.c b/tests/virtio-9p-test.c
index 5ada2839b9ae..cb086315a36e 100644
--- a/tests/virtio-9p-test.c
+++ b/tests/virtio-9p-test.c
@@ -246,13 +246,17 @@ static const char *rmessage_name(uint8_t id)
"<unknown>";
}
-static void v9fs_req_recv(P9Req *req, uint8_t id)
+static void v9fs_req_wait_for_reply(P9Req *req)
{
QVirtIO9P *v9p = req->v9p;
- P9Hdr hdr;
qvirtio_wait_used_elem(v9p->dev, v9p->vq, req->free_head,
QVIRTIO_9P_TIMEOUT_US);
+}
+
+static void v9fs_req_recv(P9Req *req, uint8_t id)
+{
+ P9Hdr hdr;
v9fs_memread(req, &hdr, 7);
hdr.size = ldl_le_p(&hdr.size);
@@ -398,6 +402,7 @@ static void fs_version(QVirtIO9P *v9p)
P9Req *req;
req = v9fs_tversion(v9p, P9_MAX_SIZE, version, P9_NOTAG);
+ v9fs_req_wait_for_reply(req);
v9fs_rversion(req, &server_len, &server_version);
g_assert_cmpmem(server_version, server_len, version, strlen(version));
@@ -411,6 +416,7 @@ static void fs_attach(QVirtIO9P *v9p)
fs_version(v9p);
req = v9fs_tattach(v9p, 0, getuid(), 0);
+ v9fs_req_wait_for_reply(req);
v9fs_rattach(req, NULL);
}
@@ -431,6 +437,7 @@ static void fs_walk(QVirtIO9P *v9p)
fs_attach(v9p);
req = v9fs_twalk(v9p, 0, 1, P9_MAXWELEM, wnames, 0);
+ v9fs_req_wait_for_reply(req);
v9fs_rwalk(req, &nwqid, &wqid);
g_assert_cmpint(nwqid, ==, P9_MAXWELEM);
@@ -452,6 +459,7 @@ static void fs_walk_no_slash(QVirtIO9P *v9p)
fs_attach(v9p);
req = v9fs_twalk(v9p, 0, 1, 1, wnames, 0);
+ v9fs_req_wait_for_reply(req);
v9fs_rlerror(req, &err);
g_assert_cmpint(err, ==, ENOENT);
@@ -467,9 +475,11 @@ static void fs_walk_dotdot(QVirtIO9P *v9p)
fs_version(v9p);
req = v9fs_tattach(v9p, 0, getuid(), 0);
+ v9fs_req_wait_for_reply(req);
v9fs_rattach(req, &root_qid);
req = v9fs_twalk(v9p, 0, 1, 1, wnames, 0);
+ v9fs_req_wait_for_reply(req);
v9fs_rwalk(req, NULL, &wqid); /* We now we'll get one qid */
g_assert_cmpmem(&root_qid, 13, wqid[0], 13);
--
2.13.6
^ permalink raw reply related [flat|nested] 13+ messages in thread* [Qemu-devel] [PULL 05/10] tests: virtio-9p: use the synth backend
2018-01-30 17:39 [Qemu-devel] [PULL 00/10] 9p patches for 2.12 20180130 Greg Kurz
` (3 preceding siblings ...)
2018-01-30 17:39 ` [Qemu-devel] [PULL 04/10] tests: virtio-9p: wait for completion in the test code Greg Kurz
@ 2018-01-30 17:39 ` Greg Kurz
2018-01-30 17:39 ` [Qemu-devel] [PULL 06/10] tests: virtio-9p: add LOPEN operation test Greg Kurz
` (5 subsequent siblings)
10 siblings, 0 replies; 13+ messages in thread
From: Greg Kurz @ 2018-01-30 17:39 UTC (permalink / raw)
To: qemu-devel; +Cc: Peter Maydell, Greg Kurz
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.
Signed-off-by: Greg Kurz <groug@kaod.org>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
---
hw/9pfs/9p-synth.c | 16 ++++++++++++++++
hw/9pfs/9p-synth.h | 4 ++++
tests/virtio-9p-test.c | 22 ++++++----------------
3 files changed, 26 insertions(+), 16 deletions(-)
diff --git a/hw/9pfs/9p-synth.c b/hw/9pfs/9p-synth.c
index 8f255e91c00f..dcbd320da17a 100644
--- a/hw/9pfs/9p-synth.c
+++ b/hw/9pfs/9p-synth.c
@@ -19,6 +19,7 @@
#include "qemu/rcu.h"
#include "qemu/rcu_queue.h"
#include "qemu/cutils.h"
+#include "sysemu/qtest.h"
/* Root node for synth file system */
static V9fsSynthNode synth_root = {
@@ -527,6 +528,21 @@ static int synth_init(FsContext *ctx, Error **errp)
/* Mark the subsystem is ready for use */
synth_fs = 1;
+
+ if (qtest_enabled()) {
+ V9fsSynthNode *node = NULL;
+ int i, ret;
+
+ /* Directory hierarchy for WALK test */
+ for (i = 0; i < P9_MAXWELEM; i++) {
+ char *name = g_strdup_printf(QTEST_V9FS_SYNTH_WALK_FILE, i);
+
+ ret = qemu_v9fs_synth_mkdir(node, 0700, name, &node);
+ assert(!ret);
+ g_free(name);
+ }
+ }
+
return 0;
}
diff --git a/hw/9pfs/9p-synth.h b/hw/9pfs/9p-synth.h
index 49c2fc7b274e..876b4ef58288 100644
--- a/hw/9pfs/9p-synth.h
+++ b/hw/9pfs/9p-synth.h
@@ -49,4 +49,8 @@ int qemu_v9fs_synth_add_file(V9fsSynthNode *parent, int mode,
const char *name, v9fs_synth_read read,
v9fs_synth_write write, void *arg);
+/* qtest stuff */
+
+#define QTEST_V9FS_SYNTH_WALK_FILE "WALK%d"
+
#endif
diff --git a/tests/virtio-9p-test.c b/tests/virtio-9p-test.c
index cb086315a36e..652198156731 100644
--- a/tests/virtio-9p-test.c
+++ b/tests/virtio-9p-test.c
@@ -17,6 +17,7 @@
#include "standard-headers/linux/virtio_ids.h"
#include "standard-headers/linux/virtio_pci.h"
#include "hw/9pfs/9p.h"
+#include "hw/9pfs/9p-synth.h"
#define QVIRTIO_9P_TIMEOUT_US (10 * 1000 * 1000)
@@ -26,23 +27,19 @@ typedef struct {
QVirtioDevice *dev;
QOSState *qs;
QVirtQueue *vq;
- char *test_share;
} QVirtIO9P;
static QVirtIO9P *qvirtio_9p_start(const char *driver)
{
const char *arch = qtest_get_arch();
- const char *cmd = "-fsdev local,id=fsdev0,security_model=none,path=%s "
+ const char *cmd = "-fsdev synth,id=fsdev0 "
"-device %s,fsdev=fsdev0,mount_tag=%s";
QVirtIO9P *v9p = g_new0(QVirtIO9P, 1);
- v9p->test_share = g_strdup("/tmp/qtest.XXXXXX");
- g_assert_nonnull(mkdtemp(v9p->test_share));
-
if (strcmp(arch, "i386") == 0 || strcmp(arch, "x86_64") == 0) {
- v9p->qs = qtest_pc_boot(cmd, v9p->test_share, driver, mount_tag);
+ v9p->qs = qtest_pc_boot(cmd, driver, mount_tag);
} else if (strcmp(arch, "ppc64") == 0) {
- v9p->qs = qtest_spapr_boot(cmd, v9p->test_share, driver, mount_tag);
+ v9p->qs = qtest_spapr_boot(cmd, driver, mount_tag);
} else {
g_printerr("virtio-9p tests are only available on x86 or ppc64\n");
exit(EXIT_FAILURE);
@@ -54,8 +51,6 @@ static QVirtIO9P *qvirtio_9p_start(const char *driver)
static void qvirtio_9p_stop(QVirtIO9P *v9p)
{
qtest_shutdown(v9p->qs);
- rmdir(v9p->test_share);
- g_free(v9p->test_share);
g_free(v9p);
}
@@ -422,17 +417,14 @@ static void fs_attach(QVirtIO9P *v9p)
static void fs_walk(QVirtIO9P *v9p)
{
- char *wnames[P9_MAXWELEM], *paths[P9_MAXWELEM];
- char *last_path = v9p->test_share;
+ char *wnames[P9_MAXWELEM];
uint16_t nwqid;
v9fs_qid *wqid;
int i;
P9Req *req;
for (i = 0; i < P9_MAXWELEM; i++) {
- wnames[i] = g_strdup_printf("%s%d", __func__, i);
- last_path = paths[i] = g_strdup_printf("%s/%s", last_path, wnames[i]);
- g_assert(!mkdir(paths[i], 0700));
+ wnames[i] = g_strdup_printf(QTEST_V9FS_SYNTH_WALK_FILE, i);
}
fs_attach(v9p);
@@ -443,8 +435,6 @@ static void fs_walk(QVirtIO9P *v9p)
g_assert_cmpint(nwqid, ==, P9_MAXWELEM);
for (i = 0; i < P9_MAXWELEM; i++) {
- rmdir(paths[P9_MAXWELEM - i - 1]);
- g_free(paths[P9_MAXWELEM - i - 1]);
g_free(wnames[i]);
}
--
2.13.6
^ permalink raw reply related [flat|nested] 13+ messages in thread* [Qemu-devel] [PULL 06/10] tests: virtio-9p: add LOPEN operation test
2018-01-30 17:39 [Qemu-devel] [PULL 00/10] 9p patches for 2.12 20180130 Greg Kurz
` (4 preceding siblings ...)
2018-01-30 17:39 ` [Qemu-devel] [PULL 05/10] tests: virtio-9p: use the synth backend Greg Kurz
@ 2018-01-30 17:39 ` Greg Kurz
2018-01-30 17:39 ` [Qemu-devel] [PULL 07/10] tests: virtio-9p: add WRITE " Greg Kurz
` (4 subsequent siblings)
10 siblings, 0 replies; 13+ messages in thread
From: Greg Kurz @ 2018-01-30 17:39 UTC (permalink / raw)
To: qemu-devel; +Cc: Peter Maydell, Greg Kurz
Trivial test of a successful open.
Signed-off-by: Greg Kurz <groug@kaod.org>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
---
hw/9pfs/9p-synth.c | 5 +++++
hw/9pfs/9p-synth.h | 1 +
tests/virtio-9p-test.c | 47 +++++++++++++++++++++++++++++++++++++++++++++++
3 files changed, 53 insertions(+)
diff --git a/hw/9pfs/9p-synth.c b/hw/9pfs/9p-synth.c
index dcbd320da17a..f17b74f44461 100644
--- a/hw/9pfs/9p-synth.c
+++ b/hw/9pfs/9p-synth.c
@@ -541,6 +541,11 @@ static int synth_init(FsContext *ctx, Error **errp)
assert(!ret);
g_free(name);
}
+
+ /* File for LOPEN test */
+ ret = qemu_v9fs_synth_add_file(NULL, 0, QTEST_V9FS_SYNTH_LOPEN_FILE,
+ NULL, NULL, ctx);
+ assert(!ret);
}
return 0;
diff --git a/hw/9pfs/9p-synth.h b/hw/9pfs/9p-synth.h
index 876b4ef58288..2a8d6fd00d69 100644
--- a/hw/9pfs/9p-synth.h
+++ b/hw/9pfs/9p-synth.h
@@ -52,5 +52,6 @@ int qemu_v9fs_synth_add_file(V9fsSynthNode *parent, int mode,
/* qtest stuff */
#define QTEST_V9FS_SYNTH_WALK_FILE "WALK%d"
+#define QTEST_V9FS_SYNTH_LOPEN_FILE "LOPEN"
#endif
diff --git a/tests/virtio-9p-test.c b/tests/virtio-9p-test.c
index 652198156731..6ba782e24f3a 100644
--- a/tests/virtio-9p-test.c
+++ b/tests/virtio-9p-test.c
@@ -238,6 +238,7 @@ static const char *rmessage_name(uint8_t id)
id == P9_RVERSION ? "RVERSION" :
id == P9_RATTACH ? "RATTACH" :
id == P9_RWALK ? "RWALK" :
+ id == P9_RLOPEN ? "RLOPEN" :
"<unknown>";
}
@@ -389,6 +390,34 @@ static void v9fs_rwalk(P9Req *req, uint16_t *nwqid, v9fs_qid **wqid)
v9fs_req_free(req);
}
+/* size[4] Tlopen tag[2] fid[4] flags[4] */
+static P9Req *v9fs_tlopen(QVirtIO9P *v9p, uint32_t fid, uint32_t flags,
+ uint16_t tag)
+{
+ P9Req *req;
+
+ req = v9fs_req_init(v9p, 4 + 4, P9_TLOPEN, tag);
+ v9fs_uint32_write(req, fid);
+ v9fs_uint32_write(req, flags);
+ v9fs_req_send(req);
+ return req;
+}
+
+/* size[4] Rlopen tag[2] qid[13] iounit[4] */
+static void v9fs_rlopen(P9Req *req, v9fs_qid *qid, uint32_t *iounit)
+{
+ v9fs_req_recv(req, P9_RLOPEN);
+ if (qid) {
+ v9fs_memread(req, qid, 13);
+ } else {
+ v9fs_memskip(req, 13);
+ }
+ if (iounit) {
+ v9fs_uint32_read(req, iounit);
+ }
+ v9fs_req_free(req);
+}
+
static void fs_version(QVirtIO9P *v9p)
{
const char *version = "9P2000.L";
@@ -478,6 +507,23 @@ static void fs_walk_dotdot(QVirtIO9P *v9p)
g_free(wnames[0]);
}
+static void fs_lopen(QVirtIO9P *v9p)
+{
+ char *const wnames[] = { g_strdup(QTEST_V9FS_SYNTH_LOPEN_FILE) };
+ P9Req *req;
+
+ fs_attach(v9p);
+ req = v9fs_twalk(v9p, 0, 1, 1, wnames, 0);
+ v9fs_req_wait_for_reply(req);
+ v9fs_rwalk(req, NULL, NULL);
+
+ req = v9fs_tlopen(v9p, 1, O_WRONLY, 0);
+ v9fs_req_wait_for_reply(req);
+ v9fs_rlopen(req, NULL, NULL);
+
+ g_free(wnames[0]);
+}
+
typedef void (*v9fs_test_fn)(QVirtIO9P *v9p);
static void v9fs_run_pci_test(gconstpointer data)
@@ -507,6 +553,7 @@ int main(int argc, char **argv)
v9fs_qtest_pci_add("/virtio/9p/pci/fs/walk/no_slash", fs_walk_no_slash);
v9fs_qtest_pci_add("/virtio/9p/pci/fs/walk/dotdot_from_root",
fs_walk_dotdot);
+ v9fs_qtest_pci_add("/virtio/9p/pci/fs/lopen/basic", fs_lopen);
return g_test_run();
}
--
2.13.6
^ permalink raw reply related [flat|nested] 13+ messages in thread* [Qemu-devel] [PULL 07/10] tests: virtio-9p: add WRITE operation test
2018-01-30 17:39 [Qemu-devel] [PULL 00/10] 9p patches for 2.12 20180130 Greg Kurz
` (5 preceding siblings ...)
2018-01-30 17:39 ` [Qemu-devel] [PULL 06/10] tests: virtio-9p: add LOPEN operation test Greg Kurz
@ 2018-01-30 17:39 ` Greg Kurz
2018-01-30 17:39 ` [Qemu-devel] [PULL 08/10] libqos/virtio: return length written into used descriptor Greg Kurz
` (3 subsequent siblings)
10 siblings, 0 replies; 13+ messages in thread
From: Greg Kurz @ 2018-01-30 17:39 UTC (permalink / raw)
To: qemu-devel; +Cc: Peter Maydell, Greg Kurz
Trivial test of a successful write.
Signed-off-by: Greg Kurz <groug@kaod.org>
(groug, handle potential overflow when computing request size)
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
---
hw/9pfs/9p-synth.c | 11 +++++++++
hw/9pfs/9p-synth.h | 1 +
tests/virtio-9p-test.c | 62 ++++++++++++++++++++++++++++++++++++++++++++++++++
3 files changed, 74 insertions(+)
diff --git a/hw/9pfs/9p-synth.c b/hw/9pfs/9p-synth.c
index f17b74f44461..f2d59a90a670 100644
--- a/hw/9pfs/9p-synth.c
+++ b/hw/9pfs/9p-synth.c
@@ -515,6 +515,12 @@ static int synth_unlinkat(FsContext *ctx, V9fsPath *dir,
return -1;
}
+static ssize_t v9fs_synth_qtest_write(void *buf, int len, off_t offset,
+ void *arg)
+{
+ return len;
+}
+
static int synth_init(FsContext *ctx, Error **errp)
{
QLIST_INIT(&synth_root.child);
@@ -546,6 +552,11 @@ static int synth_init(FsContext *ctx, Error **errp)
ret = qemu_v9fs_synth_add_file(NULL, 0, QTEST_V9FS_SYNTH_LOPEN_FILE,
NULL, NULL, ctx);
assert(!ret);
+
+ /* File for WRITE test */
+ ret = qemu_v9fs_synth_add_file(NULL, 0, QTEST_V9FS_SYNTH_WRITE_FILE,
+ NULL, v9fs_synth_qtest_write, ctx);
+ assert(!ret);
}
return 0;
diff --git a/hw/9pfs/9p-synth.h b/hw/9pfs/9p-synth.h
index 2a8d6fd00d69..a74032d7bd9a 100644
--- a/hw/9pfs/9p-synth.h
+++ b/hw/9pfs/9p-synth.h
@@ -53,5 +53,6 @@ int qemu_v9fs_synth_add_file(V9fsSynthNode *parent, int mode,
#define QTEST_V9FS_SYNTH_WALK_FILE "WALK%d"
#define QTEST_V9FS_SYNTH_LOPEN_FILE "LOPEN"
+#define QTEST_V9FS_SYNTH_WRITE_FILE "WRITE"
#endif
diff --git a/tests/virtio-9p-test.c b/tests/virtio-9p-test.c
index 6ba782e24f3a..37a7dae3f78e 100644
--- a/tests/virtio-9p-test.c
+++ b/tests/virtio-9p-test.c
@@ -150,6 +150,13 @@ static void v9fs_uint32_write(P9Req *req, uint32_t val)
v9fs_memwrite(req, &le_val, 4);
}
+static void v9fs_uint64_write(P9Req *req, uint64_t val)
+{
+ uint64_t le_val = cpu_to_le64(val);
+
+ v9fs_memwrite(req, &le_val, 8);
+}
+
static void v9fs_uint32_read(P9Req *req, uint32_t *val)
{
v9fs_memread(req, val, 4);
@@ -239,6 +246,7 @@ static const char *rmessage_name(uint8_t id)
id == P9_RATTACH ? "RATTACH" :
id == P9_RWALK ? "RWALK" :
id == P9_RLOPEN ? "RLOPEN" :
+ id == P9_RWRITE ? "RWRITE" :
"<unknown>";
}
@@ -418,6 +426,34 @@ static void v9fs_rlopen(P9Req *req, v9fs_qid *qid, uint32_t *iounit)
v9fs_req_free(req);
}
+/* size[4] Twrite tag[2] fid[4] offset[8] count[4] data[count] */
+static P9Req *v9fs_twrite(QVirtIO9P *v9p, uint32_t fid, uint64_t offset,
+ uint32_t count, const void *data, uint16_t tag)
+{
+ P9Req *req;
+ uint32_t body_size = 4 + 8 + 4;
+
+ g_assert_cmpint(body_size, <=, UINT32_MAX - count);
+ body_size += count;
+ req = v9fs_req_init(v9p, body_size, P9_TWRITE, tag);
+ v9fs_uint32_write(req, fid);
+ v9fs_uint64_write(req, offset);
+ v9fs_uint32_write(req, count);
+ v9fs_memwrite(req, data, count);
+ v9fs_req_send(req);
+ return req;
+}
+
+/* size[4] Rwrite tag[2] count[4] */
+static void v9fs_rwrite(P9Req *req, uint32_t *count)
+{
+ v9fs_req_recv(req, P9_RWRITE);
+ if (count) {
+ v9fs_uint32_read(req, count);
+ }
+ v9fs_req_free(req);
+}
+
static void fs_version(QVirtIO9P *v9p)
{
const char *version = "9P2000.L";
@@ -524,6 +560,31 @@ static void fs_lopen(QVirtIO9P *v9p)
g_free(wnames[0]);
}
+static void fs_write(QVirtIO9P *v9p)
+{
+ static const uint32_t write_count = P9_MAX_SIZE / 2;
+ char *const wnames[] = { g_strdup(QTEST_V9FS_SYNTH_WRITE_FILE) };
+ char *buf = g_malloc(write_count);
+ uint32_t count;
+ P9Req *req;
+
+ fs_attach(v9p);
+ req = v9fs_twalk(v9p, 0, 1, 1, wnames, 0);
+ v9fs_req_wait_for_reply(req);
+ v9fs_rwalk(req, NULL, NULL);
+
+ req = v9fs_tlopen(v9p, 1, O_WRONLY, 0);
+ v9fs_req_wait_for_reply(req);
+ v9fs_rlopen(req, NULL, NULL);
+
+ req = v9fs_twrite(v9p, 1, 0, write_count, buf, 0);
+ v9fs_req_wait_for_reply(req);
+ v9fs_rwrite(req, &count);
+ g_assert_cmpint(count, ==, write_count);
+
+ g_free(wnames[0]);
+}
+
typedef void (*v9fs_test_fn)(QVirtIO9P *v9p);
static void v9fs_run_pci_test(gconstpointer data)
@@ -554,6 +615,7 @@ int main(int argc, char **argv)
v9fs_qtest_pci_add("/virtio/9p/pci/fs/walk/dotdot_from_root",
fs_walk_dotdot);
v9fs_qtest_pci_add("/virtio/9p/pci/fs/lopen/basic", fs_lopen);
+ v9fs_qtest_pci_add("/virtio/9p/pci/fs/write/basic", fs_write);
return g_test_run();
}
--
2.13.6
^ permalink raw reply related [flat|nested] 13+ messages in thread* [Qemu-devel] [PULL 08/10] libqos/virtio: return length written into used descriptor
2018-01-30 17:39 [Qemu-devel] [PULL 00/10] 9p patches for 2.12 20180130 Greg Kurz
` (6 preceding siblings ...)
2018-01-30 17:39 ` [Qemu-devel] [PULL 07/10] tests: virtio-9p: add WRITE " Greg Kurz
@ 2018-01-30 17:39 ` Greg Kurz
2018-01-30 17:39 ` [Qemu-devel] [PULL 09/10] tests: virtio-9p: add FLUSH operation test Greg Kurz
` (2 subsequent siblings)
10 siblings, 0 replies; 13+ messages in thread
From: Greg Kurz @ 2018-01-30 17:39 UTC (permalink / raw)
To: qemu-devel; +Cc: Peter Maydell, Greg Kurz
When a 9p request is flushed (ie, cancelled) by the guest, the device
is expected to simply mark the request as used, without sending a 9p
reply (ie, without writing anything into the used buffer).
To be able to test this, we need access to the length written by the
device into the used descriptor. This patch adds a uint32_t * argument
to qvirtqueue_get_buf() and qvirtio_wait_used_elem() for this purpose.
All existing users are updated accordingly.
Signed-off-by: Greg Kurz <groug@kaod.org>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
---
tests/libqos/virtio.c | 25 +++++++++++++++++--------
tests/libqos/virtio.h | 3 ++-
tests/virtio-9p-test.c | 2 +-
tests/virtio-blk-test.c | 24 +++++++++++++-----------
tests/virtio-net-test.c | 6 +++---
tests/virtio-scsi-test.c | 3 ++-
6 files changed, 38 insertions(+), 25 deletions(-)
diff --git a/tests/libqos/virtio.c b/tests/libqos/virtio.c
index 0879a621c8af..0dad5c19acde 100644
--- a/tests/libqos/virtio.c
+++ b/tests/libqos/virtio.c
@@ -119,6 +119,8 @@ uint8_t qvirtio_wait_status_byte_no_isr(QVirtioDevice *d,
/*
* qvirtio_wait_used_elem:
* @desc_idx: The next expected vq->desc[] index in the used ring
+ * @len: A pointer that is filled with the length written into the buffer, may
+ * be NULL
* @timeout_us: How many microseconds to wait before failing
*
* This function waits for the next completed request on the used ring.
@@ -126,6 +128,7 @@ uint8_t qvirtio_wait_status_byte_no_isr(QVirtioDevice *d,
void qvirtio_wait_used_elem(QVirtioDevice *d,
QVirtQueue *vq,
uint32_t desc_idx,
+ uint32_t *len,
gint64 timeout_us)
{
gint64 start_time = g_get_monotonic_time();
@@ -136,7 +139,7 @@ void qvirtio_wait_used_elem(QVirtioDevice *d,
clock_step(100);
if (d->bus->get_queue_isr_status(d, vq) &&
- qvirtqueue_get_buf(vq, &got_desc_idx)) {
+ qvirtqueue_get_buf(vq, &got_desc_idx, len)) {
g_assert_cmpint(got_desc_idx, ==, desc_idx);
return;
}
@@ -304,30 +307,36 @@ void qvirtqueue_kick(QVirtioDevice *d, QVirtQueue *vq, uint32_t free_head)
/*
* qvirtqueue_get_buf:
* @desc_idx: A pointer that is filled with the vq->desc[] index, may be NULL
+ * @len: A pointer that is filled with the length written into the buffer, may
+ * be NULL
*
* This function gets the next used element if there is one ready.
*
* Returns: true if an element was ready, false otherwise
*/
-bool qvirtqueue_get_buf(QVirtQueue *vq, uint32_t *desc_idx)
+bool qvirtqueue_get_buf(QVirtQueue *vq, uint32_t *desc_idx, uint32_t *len)
{
uint16_t idx;
+ uint64_t elem_addr;
idx = readw(vq->used + offsetof(struct vring_used, idx));
if (idx == vq->last_used_idx) {
return false;
}
- if (desc_idx) {
- uint64_t elem_addr;
+ elem_addr = vq->used +
+ offsetof(struct vring_used, ring) +
+ (vq->last_used_idx % vq->size) *
+ sizeof(struct vring_used_elem);
- elem_addr = vq->used +
- offsetof(struct vring_used, ring) +
- (vq->last_used_idx % vq->size) *
- sizeof(struct vring_used_elem);
+ if (desc_idx) {
*desc_idx = readl(elem_addr + offsetof(struct vring_used_elem, id));
}
+ if (len) {
+ *len = readw(elem_addr + offsetof(struct vring_used_elem, len));
+ }
+
vq->last_used_idx++;
return true;
}
diff --git a/tests/libqos/virtio.h b/tests/libqos/virtio.h
index 0a04740adfe1..69b5b13840e7 100644
--- a/tests/libqos/virtio.h
+++ b/tests/libqos/virtio.h
@@ -124,6 +124,7 @@ uint8_t qvirtio_wait_status_byte_no_isr(QVirtioDevice *d,
void qvirtio_wait_used_elem(QVirtioDevice *d,
QVirtQueue *vq,
uint32_t desc_idx,
+ uint32_t *len,
gint64 timeout_us);
void qvirtio_wait_config_isr(QVirtioDevice *d, gint64 timeout_us);
QVirtQueue *qvirtqueue_setup(QVirtioDevice *d,
@@ -140,7 +141,7 @@ uint32_t qvirtqueue_add(QVirtQueue *vq, uint64_t data, uint32_t len, bool write,
bool next);
uint32_t qvirtqueue_add_indirect(QVirtQueue *vq, QVRingIndirectDesc *indirect);
void qvirtqueue_kick(QVirtioDevice *d, QVirtQueue *vq, uint32_t free_head);
-bool qvirtqueue_get_buf(QVirtQueue *vq, uint32_t *desc_idx);
+bool qvirtqueue_get_buf(QVirtQueue *vq, uint32_t *desc_idx, uint32_t *len);
void qvirtqueue_set_used_event(QVirtQueue *vq, uint16_t idx);
diff --git a/tests/virtio-9p-test.c b/tests/virtio-9p-test.c
index 37a7dae3f78e..cef21aea76c9 100644
--- a/tests/virtio-9p-test.c
+++ b/tests/virtio-9p-test.c
@@ -254,7 +254,7 @@ static void v9fs_req_wait_for_reply(P9Req *req)
{
QVirtIO9P *v9p = req->v9p;
- qvirtio_wait_used_elem(v9p->dev, v9p->vq, req->free_head,
+ qvirtio_wait_used_elem(v9p->dev, v9p->vq, req->free_head, NULL,
QVIRTIO_9P_TIMEOUT_US);
}
diff --git a/tests/virtio-blk-test.c b/tests/virtio-blk-test.c
index 45f368dcd966..2ac64e5e2559 100644
--- a/tests/virtio-blk-test.c
+++ b/tests/virtio-blk-test.c
@@ -193,7 +193,7 @@ static void test_basic(QVirtioDevice *dev, QGuestAllocator *alloc,
qvirtqueue_kick(dev, vq, free_head);
- qvirtio_wait_used_elem(dev, vq, free_head, QVIRTIO_BLK_TIMEOUT_US);
+ qvirtio_wait_used_elem(dev, vq, free_head, NULL, QVIRTIO_BLK_TIMEOUT_US);
status = readb(req_addr + 528);
g_assert_cmpint(status, ==, 0);
@@ -215,7 +215,7 @@ static void test_basic(QVirtioDevice *dev, QGuestAllocator *alloc,
qvirtqueue_kick(dev, vq, free_head);
- qvirtio_wait_used_elem(dev, vq, free_head, QVIRTIO_BLK_TIMEOUT_US);
+ qvirtio_wait_used_elem(dev, vq, free_head, NULL, QVIRTIO_BLK_TIMEOUT_US);
status = readb(req_addr + 528);
g_assert_cmpint(status, ==, 0);
@@ -243,7 +243,8 @@ static void test_basic(QVirtioDevice *dev, QGuestAllocator *alloc,
qvirtqueue_add(vq, req_addr + 528, 1, true, false);
qvirtqueue_kick(dev, vq, free_head);
- qvirtio_wait_used_elem(dev, vq, free_head, QVIRTIO_BLK_TIMEOUT_US);
+ qvirtio_wait_used_elem(dev, vq, free_head, NULL,
+ QVIRTIO_BLK_TIMEOUT_US);
status = readb(req_addr + 528);
g_assert_cmpint(status, ==, 0);
@@ -264,7 +265,8 @@ static void test_basic(QVirtioDevice *dev, QGuestAllocator *alloc,
qvirtqueue_kick(dev, vq, free_head);
- qvirtio_wait_used_elem(dev, vq, free_head, QVIRTIO_BLK_TIMEOUT_US);
+ qvirtio_wait_used_elem(dev, vq, free_head, NULL,
+ QVIRTIO_BLK_TIMEOUT_US);
status = readb(req_addr + 528);
g_assert_cmpint(status, ==, 0);
@@ -345,7 +347,7 @@ static void pci_indirect(void)
free_head = qvirtqueue_add_indirect(&vqpci->vq, indirect);
qvirtqueue_kick(&dev->vdev, &vqpci->vq, free_head);
- qvirtio_wait_used_elem(&dev->vdev, &vqpci->vq, free_head,
+ qvirtio_wait_used_elem(&dev->vdev, &vqpci->vq, free_head, NULL,
QVIRTIO_BLK_TIMEOUT_US);
status = readb(req_addr + 528);
g_assert_cmpint(status, ==, 0);
@@ -370,7 +372,7 @@ static void pci_indirect(void)
free_head = qvirtqueue_add_indirect(&vqpci->vq, indirect);
qvirtqueue_kick(&dev->vdev, &vqpci->vq, free_head);
- qvirtio_wait_used_elem(&dev->vdev, &vqpci->vq, free_head,
+ qvirtio_wait_used_elem(&dev->vdev, &vqpci->vq, free_head, NULL,
QVIRTIO_BLK_TIMEOUT_US);
status = readb(req_addr + 528);
g_assert_cmpint(status, ==, 0);
@@ -481,7 +483,7 @@ static void pci_msix(void)
qvirtqueue_add(&vqpci->vq, req_addr + 528, 1, true, false);
qvirtqueue_kick(&dev->vdev, &vqpci->vq, free_head);
- qvirtio_wait_used_elem(&dev->vdev, &vqpci->vq, free_head,
+ qvirtio_wait_used_elem(&dev->vdev, &vqpci->vq, free_head, NULL,
QVIRTIO_BLK_TIMEOUT_US);
status = readb(req_addr + 528);
@@ -506,7 +508,7 @@ static void pci_msix(void)
qvirtqueue_kick(&dev->vdev, &vqpci->vq, free_head);
- qvirtio_wait_used_elem(&dev->vdev, &vqpci->vq, free_head,
+ qvirtio_wait_used_elem(&dev->vdev, &vqpci->vq, free_head, NULL,
QVIRTIO_BLK_TIMEOUT_US);
status = readb(req_addr + 528);
@@ -580,7 +582,7 @@ static void pci_idx(void)
qvirtqueue_add(&vqpci->vq, req_addr + 528, 1, true, false);
qvirtqueue_kick(&dev->vdev, &vqpci->vq, free_head);
- qvirtio_wait_used_elem(&dev->vdev, &vqpci->vq, free_head,
+ qvirtio_wait_used_elem(&dev->vdev, &vqpci->vq, free_head, NULL,
QVIRTIO_BLK_TIMEOUT_US);
/* Write request */
@@ -627,9 +629,9 @@ static void pci_idx(void)
qvirtqueue_kick(&dev->vdev, &vqpci->vq, free_head);
/* We get just one notification for both requests */
- qvirtio_wait_used_elem(&dev->vdev, &vqpci->vq, write_head,
+ qvirtio_wait_used_elem(&dev->vdev, &vqpci->vq, write_head, NULL,
QVIRTIO_BLK_TIMEOUT_US);
- g_assert(qvirtqueue_get_buf(&vqpci->vq, &desc_idx));
+ g_assert(qvirtqueue_get_buf(&vqpci->vq, &desc_idx, NULL));
g_assert_cmpint(desc_idx, ==, free_head);
status = readb(req_addr + 528);
diff --git a/tests/virtio-net-test.c b/tests/virtio-net-test.c
index 635b942c3601..ea634dc05a60 100644
--- a/tests/virtio-net-test.c
+++ b/tests/virtio-net-test.c
@@ -108,7 +108,7 @@ static void rx_test(QVirtioDevice *dev,
ret = iov_send(socket, iov, 2, 0, sizeof(len) + sizeof(test));
g_assert_cmpint(ret, ==, sizeof(test) + sizeof(len));
- qvirtio_wait_used_elem(dev, vq, free_head, QVIRTIO_NET_TIMEOUT_US);
+ qvirtio_wait_used_elem(dev, vq, free_head, NULL, QVIRTIO_NET_TIMEOUT_US);
memread(req_addr + VNET_HDR_SIZE, buffer, sizeof(test));
g_assert_cmpstr(buffer, ==, "TEST");
@@ -131,7 +131,7 @@ static void tx_test(QVirtioDevice *dev,
free_head = qvirtqueue_add(vq, req_addr, 64, false, false);
qvirtqueue_kick(dev, vq, free_head);
- qvirtio_wait_used_elem(dev, vq, free_head, QVIRTIO_NET_TIMEOUT_US);
+ qvirtio_wait_used_elem(dev, vq, free_head, NULL, QVIRTIO_NET_TIMEOUT_US);
guest_free(alloc, req_addr);
ret = qemu_recv(socket, &len, sizeof(len), 0);
@@ -182,7 +182,7 @@ static void rx_stop_cont_test(QVirtioDevice *dev,
rsp = qmp("{ 'execute' : 'cont'}");
QDECREF(rsp);
- qvirtio_wait_used_elem(dev, vq, free_head, QVIRTIO_NET_TIMEOUT_US);
+ qvirtio_wait_used_elem(dev, vq, free_head, NULL, QVIRTIO_NET_TIMEOUT_US);
memread(req_addr + VNET_HDR_SIZE, buffer, sizeof(test));
g_assert_cmpstr(buffer, ==, "TEST");
diff --git a/tests/virtio-scsi-test.c b/tests/virtio-scsi-test.c
index 2934305b2bff..bcf408fbb637 100644
--- a/tests/virtio-scsi-test.c
+++ b/tests/virtio-scsi-test.c
@@ -121,7 +121,8 @@ static uint8_t virtio_scsi_do_command(QVirtIOSCSI *vs, const uint8_t *cdb,
}
qvirtqueue_kick(vs->dev, vq, free_head);
- qvirtio_wait_used_elem(vs->dev, vq, free_head, QVIRTIO_SCSI_TIMEOUT_US);
+ qvirtio_wait_used_elem(vs->dev, vq, free_head, NULL,
+ QVIRTIO_SCSI_TIMEOUT_US);
response = readb(resp_addr +
offsetof(struct virtio_scsi_cmd_resp, response));
--
2.13.6
^ permalink raw reply related [flat|nested] 13+ messages in thread* [Qemu-devel] [PULL 09/10] tests: virtio-9p: add FLUSH operation test
2018-01-30 17:39 [Qemu-devel] [PULL 00/10] 9p patches for 2.12 20180130 Greg Kurz
` (7 preceding siblings ...)
2018-01-30 17:39 ` [Qemu-devel] [PULL 08/10] libqos/virtio: return length written into used descriptor Greg Kurz
@ 2018-01-30 17:39 ` Greg Kurz
2018-01-30 17:39 ` [Qemu-devel] [PULL 10/10] tests/virtio-9p: explicitly handle potential integer overflows Greg Kurz
2018-01-31 10:14 ` [Qemu-devel] [PULL 00/10] 9p patches for 2.12 20180130 Peter Maydell
10 siblings, 0 replies; 13+ messages in thread
From: Greg Kurz @ 2018-01-30 17:39 UTC (permalink / raw)
To: qemu-devel; +Cc: Peter Maydell, Greg Kurz
The idea is to send a victim request that will possibly block in the
server and to send a flush request to cancel the victim request.
This patch adds two test to verifiy that:
- the server does not reply to a victim request that was actually
cancelled
- the server replies to the flush request after replying to the
victim request if it could not cancel it
9p request cancellation reference:
http://man.cat-v.org/plan_9/5/flush
Signed-off-by: Greg Kurz <groug@kaod.org>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
---
hw/9pfs/9p-synth.c | 24 ++++++++++
hw/9pfs/9p-synth.h | 5 +++
hw/9pfs/9p.c | 1 +
tests/virtio-9p-test.c | 117 +++++++++++++++++++++++++++++++++++++++++++------
4 files changed, 134 insertions(+), 13 deletions(-)
diff --git a/hw/9pfs/9p-synth.c b/hw/9pfs/9p-synth.c
index f2d59a90a670..0a9940dfa23c 100644
--- a/hw/9pfs/9p-synth.c
+++ b/hw/9pfs/9p-synth.c
@@ -521,6 +521,24 @@ static ssize_t v9fs_synth_qtest_write(void *buf, int len, off_t offset,
return len;
}
+static ssize_t v9fs_synth_qtest_flush_write(void *buf, int len, off_t offset,
+ void *arg)
+{
+ QtestV9fsSynthFlushData *data = buf;
+
+ assert(len == sizeof(*data));
+
+ if (data->usec_timeout) {
+ usleep(data->usec_timeout);
+
+ /* This will cause the server to call us again until we're cancelled */
+ errno = EINTR;
+ return -1;
+ }
+
+ return len;
+}
+
static int synth_init(FsContext *ctx, Error **errp)
{
QLIST_INIT(&synth_root.child);
@@ -557,6 +575,12 @@ static int synth_init(FsContext *ctx, Error **errp)
ret = qemu_v9fs_synth_add_file(NULL, 0, QTEST_V9FS_SYNTH_WRITE_FILE,
NULL, v9fs_synth_qtest_write, ctx);
assert(!ret);
+
+ /* File for FLUSH test */
+ ret = qemu_v9fs_synth_add_file(NULL, 0, QTEST_V9FS_SYNTH_FLUSH_FILE,
+ NULL, v9fs_synth_qtest_flush_write,
+ ctx);
+ assert(!ret);
}
return 0;
diff --git a/hw/9pfs/9p-synth.h b/hw/9pfs/9p-synth.h
index a74032d7bd9a..502ec6309a36 100644
--- a/hw/9pfs/9p-synth.h
+++ b/hw/9pfs/9p-synth.h
@@ -54,5 +54,10 @@ int qemu_v9fs_synth_add_file(V9fsSynthNode *parent, int mode,
#define QTEST_V9FS_SYNTH_WALK_FILE "WALK%d"
#define QTEST_V9FS_SYNTH_LOPEN_FILE "LOPEN"
#define QTEST_V9FS_SYNTH_WRITE_FILE "WRITE"
+#define QTEST_V9FS_SYNTH_FLUSH_FILE "FLUSH"
+
+typedef struct {
+ uint32_t usec_timeout;
+} QtestV9fsSynthFlushData;
#endif
diff --git a/hw/9pfs/9p.c b/hw/9pfs/9p.c
index e88bb50f1365..85a1ed8171a4 100644
--- a/hw/9pfs/9p.c
+++ b/hw/9pfs/9p.c
@@ -24,6 +24,7 @@
#include "coth.h"
#include "trace.h"
#include "migration/blocker.h"
+#include "sysemu/qtest.h"
int open_fd_hw;
int total_open_fd;
diff --git a/tests/virtio-9p-test.c b/tests/virtio-9p-test.c
index cef21aea76c9..41fa492cb778 100644
--- a/tests/virtio-9p-test.c
+++ b/tests/virtio-9p-test.c
@@ -247,14 +247,15 @@ static const char *rmessage_name(uint8_t id)
id == P9_RWALK ? "RWALK" :
id == P9_RLOPEN ? "RLOPEN" :
id == P9_RWRITE ? "RWRITE" :
+ id == P9_RFLUSH ? "RFLUSH" :
"<unknown>";
}
-static void v9fs_req_wait_for_reply(P9Req *req)
+static void v9fs_req_wait_for_reply(P9Req *req, uint32_t *len)
{
QVirtIO9P *v9p = req->v9p;
- qvirtio_wait_used_elem(v9p->dev, v9p->vq, req->free_head, NULL,
+ qvirtio_wait_used_elem(v9p->dev, v9p->vq, req->free_head, len,
QVIRTIO_9P_TIMEOUT_US);
}
@@ -454,6 +455,24 @@ static void v9fs_rwrite(P9Req *req, uint32_t *count)
v9fs_req_free(req);
}
+/* size[4] Tflush tag[2] oldtag[2] */
+static P9Req *v9fs_tflush(QVirtIO9P *v9p, uint16_t oldtag, uint16_t tag)
+{
+ P9Req *req;
+
+ req = v9fs_req_init(v9p, 2, P9_TFLUSH, tag);
+ v9fs_uint32_write(req, oldtag);
+ v9fs_req_send(req);
+ return req;
+}
+
+/* size[4] Rflush tag[2] */
+static void v9fs_rflush(P9Req *req)
+{
+ v9fs_req_recv(req, P9_RFLUSH);
+ v9fs_req_free(req);
+}
+
static void fs_version(QVirtIO9P *v9p)
{
const char *version = "9P2000.L";
@@ -462,7 +481,7 @@ static void fs_version(QVirtIO9P *v9p)
P9Req *req;
req = v9fs_tversion(v9p, P9_MAX_SIZE, version, P9_NOTAG);
- v9fs_req_wait_for_reply(req);
+ v9fs_req_wait_for_reply(req, NULL);
v9fs_rversion(req, &server_len, &server_version);
g_assert_cmpmem(server_version, server_len, version, strlen(version));
@@ -476,7 +495,7 @@ static void fs_attach(QVirtIO9P *v9p)
fs_version(v9p);
req = v9fs_tattach(v9p, 0, getuid(), 0);
- v9fs_req_wait_for_reply(req);
+ v9fs_req_wait_for_reply(req, NULL);
v9fs_rattach(req, NULL);
}
@@ -494,7 +513,7 @@ static void fs_walk(QVirtIO9P *v9p)
fs_attach(v9p);
req = v9fs_twalk(v9p, 0, 1, P9_MAXWELEM, wnames, 0);
- v9fs_req_wait_for_reply(req);
+ v9fs_req_wait_for_reply(req, NULL);
v9fs_rwalk(req, &nwqid, &wqid);
g_assert_cmpint(nwqid, ==, P9_MAXWELEM);
@@ -514,7 +533,7 @@ static void fs_walk_no_slash(QVirtIO9P *v9p)
fs_attach(v9p);
req = v9fs_twalk(v9p, 0, 1, 1, wnames, 0);
- v9fs_req_wait_for_reply(req);
+ v9fs_req_wait_for_reply(req, NULL);
v9fs_rlerror(req, &err);
g_assert_cmpint(err, ==, ENOENT);
@@ -530,11 +549,11 @@ static void fs_walk_dotdot(QVirtIO9P *v9p)
fs_version(v9p);
req = v9fs_tattach(v9p, 0, getuid(), 0);
- v9fs_req_wait_for_reply(req);
+ v9fs_req_wait_for_reply(req, NULL);
v9fs_rattach(req, &root_qid);
req = v9fs_twalk(v9p, 0, 1, 1, wnames, 0);
- v9fs_req_wait_for_reply(req);
+ v9fs_req_wait_for_reply(req, NULL);
v9fs_rwalk(req, NULL, &wqid); /* We now we'll get one qid */
g_assert_cmpmem(&root_qid, 13, wqid[0], 13);
@@ -550,11 +569,11 @@ static void fs_lopen(QVirtIO9P *v9p)
fs_attach(v9p);
req = v9fs_twalk(v9p, 0, 1, 1, wnames, 0);
- v9fs_req_wait_for_reply(req);
+ v9fs_req_wait_for_reply(req, NULL);
v9fs_rwalk(req, NULL, NULL);
req = v9fs_tlopen(v9p, 1, O_WRONLY, 0);
- v9fs_req_wait_for_reply(req);
+ v9fs_req_wait_for_reply(req, NULL);
v9fs_rlopen(req, NULL, NULL);
g_free(wnames[0]);
@@ -570,21 +589,91 @@ static void fs_write(QVirtIO9P *v9p)
fs_attach(v9p);
req = v9fs_twalk(v9p, 0, 1, 1, wnames, 0);
- v9fs_req_wait_for_reply(req);
+ v9fs_req_wait_for_reply(req, NULL);
v9fs_rwalk(req, NULL, NULL);
req = v9fs_tlopen(v9p, 1, O_WRONLY, 0);
- v9fs_req_wait_for_reply(req);
+ v9fs_req_wait_for_reply(req, NULL);
v9fs_rlopen(req, NULL, NULL);
req = v9fs_twrite(v9p, 1, 0, write_count, buf, 0);
- v9fs_req_wait_for_reply(req);
+ v9fs_req_wait_for_reply(req, NULL);
v9fs_rwrite(req, &count);
g_assert_cmpint(count, ==, write_count);
g_free(wnames[0]);
}
+static void fs_flush_success(QVirtIO9P *v9p)
+{
+ char *const wnames[] = { g_strdup(QTEST_V9FS_SYNTH_FLUSH_FILE) };
+ QtestV9fsSynthFlushData data;
+ P9Req *req, *flush_req;
+ uint32_t reply_len;
+
+ fs_attach(v9p);
+ req = v9fs_twalk(v9p, 0, 1, 1, wnames, 0);
+ v9fs_req_wait_for_reply(req, NULL);
+ v9fs_rwalk(req, NULL, NULL);
+
+ req = v9fs_tlopen(v9p, 1, O_WRONLY, 0);
+ v9fs_req_wait_for_reply(req, NULL);
+ v9fs_rlopen(req, NULL, NULL);
+
+ /* This will cause the 9p server to try to write data to the backend,
+ * until the write request gets cancelled.
+ */
+ data.usec_timeout = 10;
+ req = v9fs_twrite(v9p, 1, 0, sizeof(data), &data, 0);
+
+ flush_req = v9fs_tflush(v9p, req->tag, 1);
+
+ /* The write request is supposed to be flushed: the server should just
+ * mark the write request as used and reply to the flush request.
+ */
+ v9fs_req_wait_for_reply(req, &reply_len);
+ g_assert_cmpint(reply_len, ==, 0);
+ v9fs_req_free(req);
+ v9fs_rflush(flush_req);
+
+ g_free(wnames[0]);
+}
+
+static void fs_flush_ignored(QVirtIO9P *v9p)
+{
+ char *const wnames[] = { g_strdup(QTEST_V9FS_SYNTH_FLUSH_FILE) };
+ QtestV9fsSynthFlushData data;
+ P9Req *req, *flush_req;
+ uint32_t count;
+
+ fs_attach(v9p);
+ req = v9fs_twalk(v9p, 0, 1, 1, wnames, 0);
+ v9fs_req_wait_for_reply(req, NULL);
+ v9fs_rwalk(req, NULL, NULL);
+
+ req = v9fs_tlopen(v9p, 1, O_WRONLY, 0);
+ v9fs_req_wait_for_reply(req, NULL);
+ v9fs_rlopen(req, NULL, NULL);
+
+ /* This will cause the write request to complete right away, before it
+ * could be actually cancelled.
+ */
+ data.usec_timeout = 0;
+ req = v9fs_twrite(v9p, 1, 0, sizeof(data), &data, 0);
+
+ flush_req = v9fs_tflush(v9p, req->tag, 1);
+
+ /* The write request is supposed to complete. The server should
+ * reply to the write request and the flush request.
+ */
+ v9fs_req_wait_for_reply(req, NULL);
+ v9fs_rwrite(req, &count);
+ g_assert_cmpint(count, ==, sizeof(data));
+ v9fs_rflush(flush_req);
+
+ g_free(wnames[0]);
+}
+
typedef void (*v9fs_test_fn)(QVirtIO9P *v9p);
static void v9fs_run_pci_test(gconstpointer data)
@@ -616,6 +705,8 @@ int main(int argc, char **argv)
fs_walk_dotdot);
v9fs_qtest_pci_add("/virtio/9p/pci/fs/lopen/basic", fs_lopen);
v9fs_qtest_pci_add("/virtio/9p/pci/fs/write/basic", fs_write);
+ v9fs_qtest_pci_add("/virtio/9p/pci/fs/flush/success", fs_flush_success);
+ v9fs_qtest_pci_add("/virtio/9p/pci/fs/flush/ignored", fs_flush_ignored);
return g_test_run();
}
--
2.13.6
^ permalink raw reply related [flat|nested] 13+ messages in thread* [Qemu-devel] [PULL 10/10] tests/virtio-9p: explicitly handle potential integer overflows
2018-01-30 17:39 [Qemu-devel] [PULL 00/10] 9p patches for 2.12 20180130 Greg Kurz
` (8 preceding siblings ...)
2018-01-30 17:39 ` [Qemu-devel] [PULL 09/10] tests: virtio-9p: add FLUSH operation test Greg Kurz
@ 2018-01-30 17:39 ` Greg Kurz
2018-01-31 10:14 ` [Qemu-devel] [PULL 00/10] 9p patches for 2.12 20180130 Peter Maydell
10 siblings, 0 replies; 13+ messages in thread
From: Greg Kurz @ 2018-01-30 17:39 UTC (permalink / raw)
To: qemu-devel; +Cc: Peter Maydell, Greg Kurz
Signed-off-by: Greg Kurz <groug@kaod.org>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
---
tests/virtio-9p-test.c | 31 +++++++++++++++++++++----------
1 file changed, 21 insertions(+), 10 deletions(-)
diff --git a/tests/virtio-9p-test.c b/tests/virtio-9p-test.c
index 41fa492cb778..f4824fa33b87 100644
--- a/tests/virtio-9p-test.c
+++ b/tests/virtio-9p-test.c
@@ -168,7 +168,7 @@ static uint16_t v9fs_string_size(const char *string)
{
size_t len = strlen(string);
- g_assert_cmpint(len, <=, UINT16_MAX);
+ g_assert_cmpint(len, <=, UINT16_MAX - 2);
return 2 + len;
}
@@ -209,17 +209,20 @@ static P9Req *v9fs_req_init(QVirtIO9P *v9p, uint32_t size, uint8_t id,
uint16_t tag)
{
P9Req *req = g_new0(P9Req, 1);
- uint32_t t_size = 7 + size; /* 9P header has well-known size of 7 bytes */
+ uint32_t total_size = 7; /* 9P header has well-known size of 7 bytes */
P9Hdr hdr = {
- .size = cpu_to_le32(t_size),
.id = id,
.tag = cpu_to_le16(tag)
};
- g_assert_cmpint(t_size, <=, P9_MAX_SIZE);
+ g_assert_cmpint(total_size, <=, UINT32_MAX - size);
+ total_size += size;
+ hdr.size = cpu_to_le32(total_size);
+
+ g_assert_cmpint(total_size, <=, P9_MAX_SIZE);
req->v9p = v9p;
- req->t_size = t_size;
+ req->t_size = total_size;
req->t_msg = guest_alloc(v9p->qs->alloc, req->t_size);
v9fs_memwrite(req, &hdr, 7);
req->tag = tag;
@@ -305,8 +308,13 @@ static void v9fs_rlerror(P9Req *req, uint32_t *err)
static P9Req *v9fs_tversion(QVirtIO9P *v9p, uint32_t msize, const char *version,
uint16_t tag)
{
- P9Req *req = v9fs_req_init(v9p, 4 + v9fs_string_size(version), P9_TVERSION,
- tag);
+ P9Req *req;
+ uint32_t body_size = 4;
+ uint16_t string_size = v9fs_string_size(version);
+
+ g_assert_cmpint(body_size, <=, UINT32_MAX - string_size);
+ body_size += string_size;
+ req = v9fs_req_init(v9p, body_size, P9_TVERSION, tag);
v9fs_uint32_write(req, msize);
v9fs_string_write(req, version);
@@ -366,12 +374,15 @@ static P9Req *v9fs_twalk(QVirtIO9P *v9p, uint32_t fid, uint32_t newfid,
{
P9Req *req;
int i;
- uint32_t size = 4 + 4 + 2;
+ uint32_t body_size = 4 + 4 + 2;
for (i = 0; i < nwname; i++) {
- size += v9fs_string_size(wnames[i]);
+ uint16_t wname_size = v9fs_string_size(wnames[i]);
+
+ g_assert_cmpint(body_size, <=, UINT32_MAX - wname_size);
+ body_size += wname_size;
}
- req = v9fs_req_init(v9p, size, P9_TWALK, tag);
+ req = v9fs_req_init(v9p, body_size, P9_TWALK, tag);
v9fs_uint32_write(req, fid);
v9fs_uint32_write(req, newfid);
v9fs_uint16_write(req, nwname);
--
2.13.6
^ permalink raw reply related [flat|nested] 13+ messages in thread* Re: [Qemu-devel] [PULL 00/10] 9p patches for 2.12 20180130
2018-01-30 17:39 [Qemu-devel] [PULL 00/10] 9p patches for 2.12 20180130 Greg Kurz
` (9 preceding siblings ...)
2018-01-30 17:39 ` [Qemu-devel] [PULL 10/10] tests/virtio-9p: explicitly handle potential integer overflows Greg Kurz
@ 2018-01-31 10:14 ` Peter Maydell
2018-01-31 16:30 ` Greg Kurz
10 siblings, 1 reply; 13+ messages in thread
From: Peter Maydell @ 2018-01-31 10:14 UTC (permalink / raw)
To: Greg Kurz; +Cc: QEMU Developers
On 30 January 2018 at 17:39, Greg Kurz <groug@kaod.org> wrote:
> The following changes since commit 30d9fefe1aca1e92c785214aa9201fd7c2287d56:
>
> Merge remote-tracking branch 'remotes/kraxel/tags/input-20180129-v2-pull-request' into staging (2018-01-29 15:52:27 +0000)
>
> are available in the git repository at:
>
> https://github.com/gkurz/qemu.git tags/for-upstream
>
> for you to fetch changes up to 0dfef72861261bdfd30f2cc53d61c12c097af11a:
>
> tests/virtio-9p: explicitly handle potential integer overflows (2018-01-30 15:28:56 +0100)
>
> ----------------------------------------------------------------
> This series is mostly about 9p request cancellation. It fixes a
> long standing bug (read "specification violation") where the server
> would send an invalid response when the client has cancelled an
> in-flight request. This was causing annoying spurious EINTR returns
> in linux. The fix comes with some related testing in QTEST.
>
> Other patches are code cleanup and improvements.
>
> ----------------------------------------------------------------
> Greg Kurz (9):
> 9pfs: drop v9fs_register_transport()
> tests: virtio-9p: move request tag to the test functions
> tests: virtio-9p: wait for completion in the test code
> tests: virtio-9p: use the synth backend
> tests: virtio-9p: add LOPEN operation test
> tests: virtio-9p: add WRITE operation test
> libqos/virtio: return length written into used descriptor
> tests: virtio-9p: add FLUSH operation test
> tests/virtio-9p: explicitly handle potential integer overflows
>
> Keno Fischer (1):
> 9pfs: Correctly handle cancelled requests
Hi. This fails 'make check' on sparc hosts:
TEST: tests/virtio-9p-test... (pid=21410)
/ppc64/virtio/9p/pci/nop: OK
/ppc64/virtio/9p/pci/config: OK
/ppc64/virtio/9p/pci/fs/version/basic: OK
/ppc64/virtio/9p/pci/fs/attach/basic: OK
/ppc64/virtio/9p/pci/fs/walk/basic: OK
/ppc64/virtio/9p/pci/fs/walk/no_slash: OK
/ppc64/virtio/9p/pci/fs/walk/dotdot_from_root: OK
/ppc64/virtio/9p/pci/fs/lopen/basic: OK
/ppc64/virtio/9p/pci/fs/write/basic: OK
/ppc64/virtio/9p/pci/fs/flush/success:
Broken pipe
FAIL
GTester: last random seed: R02S3152e3829aa0155ecb8ed934af62a54f
(pid=21546)
/ppc64/virtio/9p/pci/fs/flush/ignored:
Broken pipe
FAIL
GTester: last random seed: R02Sd8d854ef25a0afc00dfe43cdab8f8bc7
(pid=21552)
FAIL: tests/virtio-9p-test
On x86 with the clang runtime sanitizer it gives errors about
misaligned loads:
/home/petmay01/linaro/qemu-for-merges/hw/9pfs/9p-synth.c:531:15:
runtime error: member access within misaligned address 0x7fb933502017
for type 'QtestV9fsSynthFlushData', which requires 4 byte alignment
0x7fb933502017: note: pointer points here
04 00 00 00 0a 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00
^
/home/petmay01/linaro/qemu-for-merges/hw/9pfs/9p-synth.c:531:15:
runtime error: load of misaligned address 0x7fb933502017 for type
'uint32_t' (aka 'unsigned int'), which requires 4 byte alignment
0x7fb933502017: note: pointer points here
04 00 00 00 0a 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00
^
/home/petmay01/linaro/qemu-for-merges/hw/9pfs/9p-synth.c:532:22:
runtime error: member access within misaligned address 0x7fb933502017
for type 'QtestV9fsSynthFlushData', which requires 4 byte alignment
0x7fb933502017: note: pointer points here
04 00 00 00 0a 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00
^
/home/petmay01/linaro/qemu-for-merges/hw/9pfs/9p-synth.c:532:22:
runtime error: load of misaligned address 0x7fb933502017 for type
'uint32_t' (aka 'unsigned int'), which requires 4 byte alignment
0x7fb933502017: note: pointer points here
04 00 00 00 0a 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00
^
/home/petmay01/linaro/qemu-for-merges/hw/9pfs/9p-synth.c:531:15:
runtime error: member access within misaligned address 0x7f33bb102017
for type 'QtestV9fsSynthFlushData', which requires 4 byte alignment
0x7f33bb102017: note: pointer points here
04 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00
^
/home/petmay01/linaro/qemu-for-merges/hw/9pfs/9p-synth.c:531:15:
runtime error: load of misaligned address 0x7f33bb102017 for type
'uint32_t' (aka 'unsigned int'), which requires 4 byte alignment
0x7f33bb102017: note: pointer points here
04 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00
^
which is probably what is causing the sparc failure.
This code looks suspicious:
static ssize_t v9fs_synth_qtest_flush_write(void *buf, int len, off_t offset,
void *arg)
{
QtestV9fsSynthFlushData *data = buf;
assert(len == sizeof(*data));
if (data->usec_timeout) {
as you can't in general cast an arbitrary pointer into a
structure and use it like that.
Given that the only thing in the data stream is a 32-bit value,
it would be simpler just to say
uint32_t usec_timeout = ldl_XX_p(buf);
I put 'XX' there because the other thing that looks weird here
is the handling of endianness. The code as written is doing
a "load a host-order 32-bit value", which would be ldl_he_p().
But should a buffer in the 9pfs code really have anything
in host-order rather than a fixed-by-protocol order?
Using ldl_le_p() or ldl_be_p() seems more plausible. (Whatever
code is generating this byte stream might also need attention.)
thanks
-- PMM
^ permalink raw reply [flat|nested] 13+ messages in thread* Re: [Qemu-devel] [PULL 00/10] 9p patches for 2.12 20180130
2018-01-31 10:14 ` [Qemu-devel] [PULL 00/10] 9p patches for 2.12 20180130 Peter Maydell
@ 2018-01-31 16:30 ` Greg Kurz
0 siblings, 0 replies; 13+ messages in thread
From: Greg Kurz @ 2018-01-31 16:30 UTC (permalink / raw)
To: Peter Maydell; +Cc: QEMU Developers
On Wed, 31 Jan 2018 10:14:46 +0000
Peter Maydell <peter.maydell@linaro.org> wrote:
> On 30 January 2018 at 17:39, Greg Kurz <groug@kaod.org> wrote:
> > The following changes since commit 30d9fefe1aca1e92c785214aa9201fd7c2287d56:
> >
> > Merge remote-tracking branch 'remotes/kraxel/tags/input-20180129-v2-pull-request' into staging (2018-01-29 15:52:27 +0000)
> >
> > are available in the git repository at:
> >
> > https://github.com/gkurz/qemu.git tags/for-upstream
> >
> > for you to fetch changes up to 0dfef72861261bdfd30f2cc53d61c12c097af11a:
> >
> > tests/virtio-9p: explicitly handle potential integer overflows (2018-01-30 15:28:56 +0100)
> >
> > ----------------------------------------------------------------
> > This series is mostly about 9p request cancellation. It fixes a
> > long standing bug (read "specification violation") where the server
> > would send an invalid response when the client has cancelled an
> > in-flight request. This was causing annoying spurious EINTR returns
> > in linux. The fix comes with some related testing in QTEST.
> >
> > Other patches are code cleanup and improvements.
> >
> > ----------------------------------------------------------------
> > Greg Kurz (9):
> > 9pfs: drop v9fs_register_transport()
> > tests: virtio-9p: move request tag to the test functions
> > tests: virtio-9p: wait for completion in the test code
> > tests: virtio-9p: use the synth backend
> > tests: virtio-9p: add LOPEN operation test
> > tests: virtio-9p: add WRITE operation test
> > libqos/virtio: return length written into used descriptor
> > tests: virtio-9p: add FLUSH operation test
> > tests/virtio-9p: explicitly handle potential integer overflows
> >
> > Keno Fischer (1):
> > 9pfs: Correctly handle cancelled requests
>
> Hi. This fails 'make check' on sparc hosts:
>
> TEST: tests/virtio-9p-test... (pid=21410)
> /ppc64/virtio/9p/pci/nop: OK
> /ppc64/virtio/9p/pci/config: OK
> /ppc64/virtio/9p/pci/fs/version/basic: OK
> /ppc64/virtio/9p/pci/fs/attach/basic: OK
> /ppc64/virtio/9p/pci/fs/walk/basic: OK
> /ppc64/virtio/9p/pci/fs/walk/no_slash: OK
> /ppc64/virtio/9p/pci/fs/walk/dotdot_from_root: OK
> /ppc64/virtio/9p/pci/fs/lopen/basic: OK
> /ppc64/virtio/9p/pci/fs/write/basic: OK
> /ppc64/virtio/9p/pci/fs/flush/success:
> Broken pipe
> FAIL
> GTester: last random seed: R02S3152e3829aa0155ecb8ed934af62a54f
> (pid=21546)
> /ppc64/virtio/9p/pci/fs/flush/ignored:
> Broken pipe
> FAIL
> GTester: last random seed: R02Sd8d854ef25a0afc00dfe43cdab8f8bc7
> (pid=21552)
> FAIL: tests/virtio-9p-test
>
> On x86 with the clang runtime sanitizer it gives errors about
> misaligned loads:
>
> /home/petmay01/linaro/qemu-for-merges/hw/9pfs/9p-synth.c:531:15:
> runtime error: member access within misaligned address 0x7fb933502017
> for type 'QtestV9fsSynthFlushData', which requires 4 byte alignment
> 0x7fb933502017: note: pointer points here
> 04 00 00 00 0a 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
> 00 00 00 00 00 00 00 00 00 00
> ^
> /home/petmay01/linaro/qemu-for-merges/hw/9pfs/9p-synth.c:531:15:
> runtime error: load of misaligned address 0x7fb933502017 for type
> 'uint32_t' (aka 'unsigned int'), which requires 4 byte alignment
> 0x7fb933502017: note: pointer points here
> 04 00 00 00 0a 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
> 00 00 00 00 00 00 00 00 00 00
> ^
> /home/petmay01/linaro/qemu-for-merges/hw/9pfs/9p-synth.c:532:22:
> runtime error: member access within misaligned address 0x7fb933502017
> for type 'QtestV9fsSynthFlushData', which requires 4 byte alignment
> 0x7fb933502017: note: pointer points here
> 04 00 00 00 0a 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
> 00 00 00 00 00 00 00 00 00 00
> ^
> /home/petmay01/linaro/qemu-for-merges/hw/9pfs/9p-synth.c:532:22:
> runtime error: load of misaligned address 0x7fb933502017 for type
> 'uint32_t' (aka 'unsigned int'), which requires 4 byte alignment
> 0x7fb933502017: note: pointer points here
> 04 00 00 00 0a 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
> 00 00 00 00 00 00 00 00 00 00
> ^
> /home/petmay01/linaro/qemu-for-merges/hw/9pfs/9p-synth.c:531:15:
> runtime error: member access within misaligned address 0x7f33bb102017
> for type 'QtestV9fsSynthFlushData', which requires 4 byte alignment
> 0x7f33bb102017: note: pointer points here
> 04 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
> 00 00 00 00 00 00 00 00 00 00
> ^
> /home/petmay01/linaro/qemu-for-merges/hw/9pfs/9p-synth.c:531:15:
> runtime error: load of misaligned address 0x7f33bb102017 for type
> 'uint32_t' (aka 'unsigned int'), which requires 4 byte alignment
> 0x7f33bb102017: note: pointer points here
> 04 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
> 00 00 00 00 00 00 00 00 00 00
> ^
>
> which is probably what is causing the sparc failure.
>
> This code looks suspicious:
>
> static ssize_t v9fs_synth_qtest_flush_write(void *buf, int len, off_t offset,
> void *arg)
> {
> QtestV9fsSynthFlushData *data = buf;
>
> assert(len == sizeof(*data));
>
> if (data->usec_timeout) {
>
> as you can't in general cast an arbitrary pointer into a
> structure and use it like that.
>
... which is likely to happen if it points to some data coming from
a QTEST originated stream of bytes... :-\
> Given that the only thing in the data stream is a 32-bit value,
> it would be simpler just to say
> uint32_t usec_timeout = ldl_XX_p(buf);
>
Well, at the present time there's only a 32-bit value but this
may change. So I'd prefer to copy the bytes into an aligned
structure.
> I put 'XX' there because the other thing that looks weird here
> is the handling of endianness. The code as written is doing
> a "load a host-order 32-bit value", which would be ldl_he_p().
> But should a buffer in the 9pfs code really have anything
> in host-order rather than a fixed-by-protocol order?
> Using ldl_le_p() or ldl_be_p() seems more plausible. (Whatever
> code is generating this byte stream might also need attention.)
>
The code at the other end is test code in QTEST, the goal of
which is to simulate a blocking 9P "write" request. Is it safe
to assume that this test code does "host-order" stores ?
> thanks
> -- PMM
^ permalink raw reply [flat|nested] 13+ messages in thread