[PULL v2 00/13] Misc fixes patches

[PULL v2 00/13] Misc fixes patches

[Daniel P. Berrangé]

The following changes since commit 99e54ab5e7a6efc945af6d5661842155d1f3fc7a:

  Merge tag 'hw-misc-20260903' of https://github.com/philmd/qemu into staging (2026-09-03 16:43:24 +0100)

are available in the Git repository at:

  https://gitlab.com/berrange/qemu tags/misc-fixes-pull-request

for you to fetch changes up to 39dfe7cf8e87d6de975bae6c10c3712f4df56ee6:

  docs/system/security: exclude uninitialized stack variables as bugs (2026-09-04 14:29:16 +0100)

----------------------------------------------------------------
Merge crypto, I/O and misc fixes

 * Deprecated the AF_ALG crypto backend
 * Improve checkpatch output in CI jobs
 * Fix --disable-containers arg handling in configure
 * Document security policy for uninitialized stack variables
 * Fix multiple denial of service flaws in websockets
 * Fix error handling in some x509 APIs

----------------------------------------------------------------

Daniel P. Berrangé (4):
  crypto: deprecate the AF_ALG crypto backend
  gitlab: use --emacs --quiet for checkpatch.pl instead of --terse
  configure: correctly honour --disable-containers
  docs/system/security: exclude uninitialized stack variables as bugs

Denis V. Lunev (6):
  io/channel-socket: do not treat a zero length write as an error
  io/channel-websock: send an HTTP 400 when the greeting has no space
  io/channel-websock: handle a blocked write during the handshake
  tests/unit: add websock handshake test
  io/channel-websock: do not lose QIO_CHANNEL_ERR_BLOCK while reading
  tests/unit: cover blocked IO during the websock handshake

Evgeny Kolmakov (1):
  crypto: Use g_autofree

Marc-André Lureau (2):
  crypto/x509-utils: don't double set errp
  crypto/x509-utils: propagate the error

 .gitlab-ci.d/check-patch.py          |   6 +-
 configure                            |   1 +
 crypto/block.c                       |  17 +-
 crypto/hmac-gcrypt.c                 |  12 +-
 crypto/hmac-glib.c                   |  10 +-
 crypto/ivgen-essiv.c                 |  15 +-
 crypto/ivgen.c                       |   6 +-
 crypto/secret_keyring.c              |   5 +-
 crypto/x509-utils.c                  |   4 +-
 docs/about/deprecated.rst            |  21 +++
 docs/system/security.rst             |  10 ++
 io/channel-socket.c                  |   2 +-
 io/channel-websock.c                 |  10 +-
 meson.build                          |   6 +
 tests/unit/meson.build               |   1 +
 tests/unit/test-io-channel-websock.c | 249 +++++++++++++++++++++++++++
 16 files changed, 327 insertions(+), 48 deletions(-)
 create mode 100644 tests/unit/test-io-channel-websock.c

-- 
2.55.0

[PULL v2 01/13] crypto: Use g_autofree

[Daniel P. Berrangé]

From: Evgeny Kolmakov <randomjack94dev@gmail.com>

Use g_autofree attribute to reduce the amount
of manual g_free() calls and 'goto out' code

Signed-off-by: Evgeny Kolmakov <randomjack94dev@gmail.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
---
 crypto/block.c          | 17 ++++++-----------
 crypto/hmac-gcrypt.c    | 12 ++++--------
 crypto/hmac-glib.c      | 10 +++-------
 crypto/ivgen-essiv.c    | 15 ++++-----------
 crypto/ivgen.c          |  6 ++----
 crypto/secret_keyring.c |  5 ++---
 6 files changed, 21 insertions(+), 44 deletions(-)

diff --git a/crypto/block.c b/crypto/block.c
index 96c83e60b9..42558f3caf 100644
--- a/crypto/block.c
+++ b/crypto/block.c
@@ -55,7 +55,7 @@ QCryptoBlock *qcrypto_block_open(QCryptoBlockOpenOptions *options,
                                  unsigned int flags,
                                  Error **errp)
 {
-    QCryptoBlock *block = g_new0(QCryptoBlock, 1);
+    g_autofree QCryptoBlock *block = g_new0(QCryptoBlock, 1);
 
     qemu_mutex_init(&block->mutex);
 
@@ -65,7 +65,6 @@ QCryptoBlock *qcrypto_block_open(QCryptoBlockOpenOptions *options,
         !qcrypto_block_drivers[options->format]) {
         error_setg(errp, "Unsupported block driver %s",
                    QCryptoBlockFormat_str(options->format));
-        g_free(block);
         return NULL;
     }
 
@@ -74,11 +73,10 @@ QCryptoBlock *qcrypto_block_open(QCryptoBlockOpenOptions *options,
     if (block->driver->open(block, options, optprefix,
                             readfunc, opaque, flags, errp) < 0)
     {
-        g_free(block);
         return NULL;
     }
 
-    return block;
+    return g_steal_pointer(&block);
 }
 
 
@@ -90,7 +88,7 @@ QCryptoBlock *qcrypto_block_create(QCryptoBlockCreateOptions *options,
                                    unsigned int flags,
                                    Error **errp)
 {
-    QCryptoBlock *block = g_new0(QCryptoBlock, 1);
+    g_autofree QCryptoBlock *block = g_new0(QCryptoBlock, 1);
 
     qemu_mutex_init(&block->mutex);
 
@@ -100,7 +98,6 @@ QCryptoBlock *qcrypto_block_create(QCryptoBlockCreateOptions *options,
         !qcrypto_block_drivers[options->format]) {
         error_setg(errp, "Unsupported block driver %s",
                    QCryptoBlockFormat_str(options->format));
-        g_free(block);
         return NULL;
     }
 
@@ -109,11 +106,10 @@ QCryptoBlock *qcrypto_block_create(QCryptoBlockCreateOptions *options,
 
     if (block->driver->create(block, options, optprefix, initfunc,
                               writefunc, opaque, errp) < 0) {
-        g_free(block);
         return NULL;
     }
 
-    return block;
+    return g_steal_pointer(&block);
 }
 
 
@@ -185,17 +181,16 @@ int qcrypto_block_amend_options(QCryptoBlock *block,
 QCryptoBlockInfo *qcrypto_block_get_info(QCryptoBlock *block,
                                          Error **errp)
 {
-    QCryptoBlockInfo *info = g_new0(QCryptoBlockInfo, 1);
+    g_autofree QCryptoBlockInfo *info = g_new0(QCryptoBlockInfo, 1);
 
     info->format = block->format;
 
     if (block->driver->get_info &&
         block->driver->get_info(block, info, errp) < 0) {
-        g_free(info);
         return NULL;
     }
 
-    return info;
+    return g_steal_pointer(&info);
 }
 
 
diff --git a/crypto/hmac-gcrypt.c b/crypto/hmac-gcrypt.c
index e428d17479..44631fb348 100644
--- a/crypto/hmac-gcrypt.c
+++ b/crypto/hmac-gcrypt.c
@@ -50,7 +50,7 @@ void *qcrypto_hmac_ctx_new(QCryptoHashAlgo alg,
                            const uint8_t *key, size_t nkey,
                            Error **errp)
 {
-    QCryptoHmacGcrypt *ctx;
+    g_autofree QCryptoHmacGcrypt *ctx = NULL;
     gcry_error_t err;
 
     if (!qcrypto_hmac_supports(alg)) {
@@ -66,7 +66,7 @@ void *qcrypto_hmac_ctx_new(QCryptoHashAlgo alg,
     if (err != 0) {
         error_setg(errp, "Cannot initialize hmac: %s",
                    gcry_strerror(err));
-        goto error;
+        return NULL;
     }
 
     err = gcry_mac_setkey(ctx->handle, (const void *)key, nkey);
@@ -74,14 +74,10 @@ void *qcrypto_hmac_ctx_new(QCryptoHashAlgo alg,
         error_setg(errp, "Cannot set key: %s",
                    gcry_strerror(err));
         gcry_mac_close(ctx->handle);
-        goto error;
+        return NULL;
     }
 
-    return ctx;
-
-error:
-    g_free(ctx);
-    return NULL;
+    return g_steal_pointer(&ctx);
 }
 
 static void
diff --git a/crypto/hmac-glib.c b/crypto/hmac-glib.c
index b845133a05..1f17769c1c 100644
--- a/crypto/hmac-glib.c
+++ b/crypto/hmac-glib.c
@@ -46,7 +46,7 @@ void *qcrypto_hmac_ctx_new(QCryptoHashAlgo alg,
                            const uint8_t *key, size_t nkey,
                            Error **errp)
 {
-    QCryptoHmacGlib *ctx;
+    g_autofree QCryptoHmacGlib *ctx = NULL;
 
     if (!qcrypto_hmac_supports(alg)) {
         error_setg(errp, "Unsupported hmac algorithm %s",
@@ -60,14 +60,10 @@ void *qcrypto_hmac_ctx_new(QCryptoHashAlgo alg,
                             (const uint8_t *)key, nkey);
     if (!ctx->ghmac) {
         error_setg(errp, "Cannot initialize hmac and set key");
-        goto error;
+        return NULL;
     }
 
-    return ctx;
-
-error:
-    g_free(ctx);
-    return NULL;
+    return g_steal_pointer(&ctx);
 }
 
 static void
diff --git a/crypto/ivgen-essiv.c b/crypto/ivgen-essiv.c
index 3d5a188795..d5fa269888 100644
--- a/crypto/ivgen-essiv.c
+++ b/crypto/ivgen-essiv.c
@@ -31,10 +31,10 @@ static int qcrypto_ivgen_essiv_init(QCryptoIVGen *ivgen,
                                     const uint8_t *key, size_t nkey,
                                     Error **errp)
 {
-    uint8_t *salt;
+    g_autofree uint8_t *salt = NULL;
     size_t nhash;
     size_t nsalt;
-    QCryptoIVGenESSIV *essiv = g_new0(QCryptoIVGenESSIV, 1);
+    g_autofree QCryptoIVGenESSIV *essiv = g_new0(QCryptoIVGenESSIV, 1);
 
     /* Not necessarily the same as nkey */
     nsalt = qcrypto_cipher_get_key_len(ivgen->cipher);
@@ -46,8 +46,6 @@ static int qcrypto_ivgen_essiv_init(QCryptoIVGen *ivgen,
     if (qcrypto_hash_bytes(ivgen->hash, (const gchar *)key, nkey,
                            &salt, &nhash,
                            errp) < 0) {
-        g_free(essiv);
-        g_free(salt);
         return -1;
     }
 
@@ -57,13 +55,10 @@ static int qcrypto_ivgen_essiv_init(QCryptoIVGen *ivgen,
                                        salt, MIN(nhash, nsalt),
                                        errp);
     if (!essiv->cipher) {
-        g_free(essiv);
-        g_free(salt);
         return -1;
     }
 
-    g_free(salt);
-    ivgen->private = essiv;
+    ivgen->private = g_steal_pointer(&essiv);
 
     return 0;
 }
@@ -75,7 +70,7 @@ static int qcrypto_ivgen_essiv_calculate(QCryptoIVGen *ivgen,
 {
     QCryptoIVGenESSIV *essiv = ivgen->private;
     size_t ndata = qcrypto_cipher_get_block_len(ivgen->cipher);
-    uint8_t *data = g_new(uint8_t, ndata);
+    g_autofree uint8_t *data = g_new(uint8_t, ndata);
 
     sector = cpu_to_le64(sector);
     memcpy(data, (uint8_t *)&sector, MIN(sizeof(sector), ndata));
@@ -88,7 +83,6 @@ static int qcrypto_ivgen_essiv_calculate(QCryptoIVGen *ivgen,
                                data,
                                ndata,
                                errp) < 0) {
-        g_free(data);
         return -1;
     }
 
@@ -99,7 +93,6 @@ static int qcrypto_ivgen_essiv_calculate(QCryptoIVGen *ivgen,
     if (ndata < niv) {
         memset(iv + ndata, 0, niv - ndata);
     }
-    g_free(data);
     return 0;
 }
 
diff --git a/crypto/ivgen.c b/crypto/ivgen.c
index 6b7d24d889..9f1f7d7dca 100644
--- a/crypto/ivgen.c
+++ b/crypto/ivgen.c
@@ -33,7 +33,7 @@ QCryptoIVGen *qcrypto_ivgen_new(QCryptoIVGenAlgo alg,
                                 const uint8_t *key, size_t nkey,
                                 Error **errp)
 {
-    QCryptoIVGen *ivgen = g_new0(QCryptoIVGen, 1);
+    g_autofree QCryptoIVGen *ivgen = g_new0(QCryptoIVGen, 1);
 
     ivgen->algorithm = alg;
     ivgen->cipher = cipheralg;
@@ -51,16 +51,14 @@ QCryptoIVGen *qcrypto_ivgen_new(QCryptoIVGenAlgo alg,
         break;
     default:
         error_setg(errp, "Unknown block IV generator algorithm %d", alg);
-        g_free(ivgen);
         return NULL;
     }
 
     if (ivgen->driver->init(ivgen, key, nkey, errp) < 0) {
-        g_free(ivgen);
         return NULL;
     }
 
-    return ivgen;
+    return g_steal_pointer(&ivgen);
 }
 
 
diff --git a/crypto/secret_keyring.c b/crypto/secret_keyring.c
index 78d7f09b3b..3b332276ef 100644
--- a/crypto/secret_keyring.c
+++ b/crypto/secret_keyring.c
@@ -41,7 +41,7 @@ qcrypto_secret_keyring_load_data(QCryptoSecretCommon *sec_common,
                                  Error **errp)
 {
     QCryptoSecretKeyring *secret = QCRYPTO_SECRET_KEYRING(sec_common);
-    uint8_t *buffer = NULL;
+    g_autofree uint8_t *buffer = NULL;
     long retcode;
 
     *output = NULL;
@@ -61,12 +61,11 @@ qcrypto_secret_keyring_load_data(QCryptoSecretCommon *sec_common,
 
     retcode = keyctl_read(secret->serial, buffer, retcode);
     if (retcode < 0) {
-        g_free(buffer);
         goto keyctl_error;
     }
 
     *outputlen = retcode;
-    *output = buffer;
+    *output = g_steal_pointer(&buffer);
     return;
 
 keyctl_error:
-- 
2.55.0

[PULL v2 02/13] crypto/x509-utils: don't double set errp

[Daniel P. Berrangé]

From: Marc-André Lureau <marcandre.lureau@redhat.com>

qcrypto_x509_get_ecc_curve() already sets errp on failure.

Fixes: e8317c4c9f68 ("crypto/x509-utils: Add helper functions for DIAG 320 subcode 2")
Reviewed-by: Zhuoying Cai <zycai@linux.ibm.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
---
 crypto/x509-utils.c | 1 -
 1 file changed, 1 deletion(-)

diff --git a/crypto/x509-utils.c b/crypto/x509-utils.c
index e4767f9838..34cbfca26b 100644
--- a/crypto/x509-utils.c
+++ b/crypto/x509-utils.c
@@ -325,7 +325,6 @@ int qcrypto_x509_check_ecc_curve_p521(uint8_t *cert, size_t size, Error **errp)
 
     curve_id = qcrypto_x509_get_ecc_curve(cert, size, errp);
     if (curve_id == -1) {
-        error_setg(errp, "Failed to get ECC curve");
         return -1;
     }
 
-- 
2.55.0

[PULL v2 03/13] crypto/x509-utils: propagate the error

[Daniel P. Berrangé]

From: Marc-André Lureau <marcandre.lureau@redhat.com>

Propagate the error when qcrypto_x509_get_pk_algorithm fails, instead of
falling through to return 0 (success) with *errp set.

Fixes: e8317c4c9f68 ("crypto/x509-utils: Add helper functions for DIAG 320 subcode 2")
Reviewed-by: Zhuoying Cai <zycai@linux.ibm.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
---
 crypto/x509-utils.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/crypto/x509-utils.c b/crypto/x509-utils.c
index 34cbfca26b..edcc44de80 100644
--- a/crypto/x509-utils.c
+++ b/crypto/x509-utils.c
@@ -319,6 +319,9 @@ int qcrypto_x509_check_ecc_curve_p521(uint8_t *cert, size_t size, Error **errp)
     int curve_id;
 
     algo = qcrypto_x509_get_pk_algorithm(cert, size, errp);
+    if (algo < 0) {
+        return -1;
+    }
     if (algo != GNUTLS_PK_ECDSA) {
         return 0;
     }
-- 
2.55.0

[PULL v2 04/13] io/channel-socket: do not treat a zero length write as an error

[Daniel P. Berrangé]

From: Denis V. Lunev <den@openvz.org>

qio_channel_socket_writev() checks "ret <= 0" after sendmsg(). A zero
length iovec is written successfully and returns 0, so the success
falls into the errno switch, which acts on whatever the last failing
syscall left in errno. A stale EAGAIN turns it into
QIO_CHANNEL_ERR_BLOCK with errp untouched, and a caller which treats
every negative return as fatal then passes a NULL Error to
error_get_pretty(). The websocket handshake does exactly that, so an
unauthenticated client crashes QEMU during the greeting.

Returning 0 is safe for callers which loop until everything is
written. qio_channel_writev_full_all() has no zero progress guard, but
iov_copy() yields no entries for a zero length write, so that loop is
never entered. A connected stream socket returns 0 only when there is
nothing to send.

The WIN32 implementation in the same file uses "ret < 0".

Fixes: CVE-2026-84788
Fixes: 559607ea173a ("io: add QIOChannelSocket class")
Cc: qemu-stable@nongnu.org
Cc: Daniel P. Berrangé <berrange@redhat.com>
Cc: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Denis V. Lunev <den@openvz.org>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
---
 io/channel-socket.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/io/channel-socket.c b/io/channel-socket.c
index 12773b832c..7920cee639 100644
--- a/io/channel-socket.c
+++ b/io/channel-socket.c
@@ -667,7 +667,7 @@ static ssize_t qio_channel_socket_writev(QIOChannel *ioc,
 
  retry:
     ret = sendmsg(sioc->fd, &msg, sflags);
-    if (ret <= 0) {
+    if (ret < 0) {
         switch (errno) {
         case EAGAIN:
             return QIO_CHANNEL_ERR_BLOCK;
-- 
2.55.0

[PULL v2 05/13] io/channel-websock: send an HTTP 400 when the greeting has no space

[Daniel P. Berrangé]

From: Denis V. Lunev <den@openvz.org>

qio_channel_websock_extract_headers() returns 0 without queueing a
response when the request line contains no space, unlike every sibling
check which jumps to bad_request. encoutput stays empty, yet
qio_channel_websock_handshake_read() still reports success and the
caller arms a G_IO_OUT watch to flush nothing.

Flushing that empty buffer is where QEMU crashes. Any client can
trigger it before authentication on a VNC websocket port:

  printf 'stats\r\nx\r\n\r\n' | nc $host $port

Fixes: 07e95cd529af ("io: fully parse & validate HTTP headers for websocket protocol handshake")
Fixes: f69a8bde2935 ("io: send proper HTTP response for websocket errors")
Fixes: CVE-2026-84788
Cc: qemu-stable@nongnu.org
Cc: Daniel P. Berrangé <berrange@redhat.com>
Cc: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Denis V. Lunev <den@openvz.org>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
---
 io/channel-websock.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/io/channel-websock.c b/io/channel-websock.c
index 1929abf56a..66c91ed2a2 100644
--- a/io/channel-websock.c
+++ b/io/channel-websock.c
@@ -230,7 +230,7 @@ qio_channel_websock_extract_headers(QIOChannelWebsock *ioc,
     tmp = strchr(buffer, ' ');
     if (!tmp) {
         error_setg(errp, "Missing HTTP path delimiter");
-        return 0;
+        goto bad_request;
     }
     *tmp = '\0';
 
-- 
2.55.0

[PULL v2 06/13] io/channel-websock: handle a blocked write during the handshake

[Daniel P. Berrangé]

From: Denis V. Lunev <den@openvz.org>

qio_channel_websock_handshake_send() treats every negative return from
qio_channel_write() as fatal and passes err to error_get_pretty().
QIO_CHANNEL_ERR_BLOCK is negative but leaves err NULL, so a socket
which cannot take the response immediately crashes QEMU before the
client has authenticated.

Keep the G_IO_OUT watch armed and retry instead.

Fixes: 2d1d0e70cf3e ("io: add QIOChannelWebsock class")
Fixes: CVE-2026-84788
Cc: qemu-stable@nongnu.org
Cc: Daniel P. Berrangé <berrange@redhat.com>
Cc: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Denis V. Lunev <den@openvz.org>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
---
 io/channel-websock.c | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/io/channel-websock.c b/io/channel-websock.c
index 66c91ed2a2..8f27b1f12b 100644
--- a/io/channel-websock.c
+++ b/io/channel-websock.c
@@ -562,6 +562,11 @@ static gboolean qio_channel_websock_handshake_send(QIOChannel *ioc,
                             wioc->encoutput.offset,
                             &err);
 
+    if (ret == QIO_CHANNEL_ERR_BLOCK) {
+        /* Socket buffer is full, the G_IO_OUT watch stays armed */
+        return TRUE;
+    }
+
     if (ret < 0) {
         trace_qio_channel_websock_handshake_fail(ioc, error_get_pretty(err));
         qio_task_set_error(task, err);
-- 
2.55.0

[PULL v2 07/13] tests/unit: add websock handshake test

[Daniel P. Berrangé]

From: Denis V. Lunev <den@openvz.org>

Check that malformed HTTP greetings are answered with an HTTP 400
rather than an empty response. The no-space case is the one which used
to leave the response buffer empty.

Fixes: CVE-2026-84788
Cc: Daniel P. Berrangé <berrange@redhat.com>
Cc: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Denis V. Lunev <den@openvz.org>
[DB: exclude test from Windows since it depends on AF_UNIX
 which is not universally available]
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
---
 tests/unit/meson.build               |   1 +
 tests/unit/test-io-channel-websock.c | 105 +++++++++++++++++++++++++++
 2 files changed, 106 insertions(+)
 create mode 100644 tests/unit/test-io-channel-websock.c

diff --git a/tests/unit/meson.build b/tests/unit/meson.build
index 3a9866c1f2..e47bc7225a 100644
--- a/tests/unit/meson.build
+++ b/tests/unit/meson.build
@@ -115,6 +115,7 @@ if have_block
   endif
   if host_os != 'windows'
     tests += {
+      'test-io-channel-websock': [io],
       'test-image-locking': [testblock],
       'test-nested-aio-poll': [],
     }
diff --git a/tests/unit/test-io-channel-websock.c b/tests/unit/test-io-channel-websock.c
new file mode 100644
index 0000000000..2a55a4bcdf
--- /dev/null
+++ b/tests/unit/test-io-channel-websock.c
@@ -0,0 +1,105 @@
+/*
+ * SPDX-License-Identifier: GPL-2.0-or-later
+ *
+ * QEMU I/O channel websock test
+ *
+ * Copyright (c) 2026 Virtuozzo International GmbH
+ */
+
+#include "qemu/osdep.h"
+#include "io/channel-websock.h"
+#include "io/channel-socket.h"
+#include "qapi/error.h"
+#include "qemu/module.h"
+#include "qemu/sockets.h"
+
+typedef struct {
+    bool finished;
+    bool failed;
+} QIOChannelWebsockHandshake;
+
+static void test_websock_handshake_done(QIOTask *task, gpointer opaque)
+{
+    QIOChannelWebsockHandshake *res = opaque;
+
+    res->finished = true;
+    res->failed = qio_task_propagate_error(task, NULL);
+}
+
+/*
+ * Drives a server-side handshake against @request and returns whatever
+ * the server wrote back, NUL terminated. The handshake is expected to
+ * fail; the point of the test is the HTTP response that goes with it.
+ */
+static char *test_websock_handshake_reply(const char *request)
+{
+    QIOChannelWebsockHandshake res = { false, false };
+    QIOChannelSocket *cli, *srv;
+    QIOChannelWebsock *wioc;
+    GMainContext *mainloop;
+    int channel[2];
+    char *reply;
+    ssize_t got;
+
+    g_assert(qemu_socketpair(AF_UNIX, SOCK_STREAM, 0, channel) == 0);
+
+    cli = qio_channel_socket_new_fd(channel[0], &error_abort);
+    srv = qio_channel_socket_new_fd(channel[1], &error_abort);
+    qio_channel_set_blocking(QIO_CHANNEL(srv), false, &error_abort);
+    qio_channel_set_blocking(QIO_CHANNEL(cli), false, &error_abort);
+
+    wioc = qio_channel_websock_new_server(QIO_CHANNEL(srv));
+    qio_channel_websock_handshake(wioc, test_websock_handshake_done,
+                                  &res, NULL);
+
+    qio_channel_write_all(QIO_CHANNEL(cli), request, strlen(request),
+                          &error_abort);
+
+    mainloop = g_main_context_default();
+    while (!res.finished) {
+        g_main_context_iteration(mainloop, TRUE);
+    }
+    g_assert(res.failed);
+
+    reply = g_malloc0(1024);
+    got = qio_channel_read(QIO_CHANNEL(cli), reply, 1023, &error_abort);
+    if (got > 0) {
+        reply[got] = '\0';
+    }
+
+    object_unref(OBJECT(wioc));
+    object_unref(OBJECT(srv));
+    object_unref(OBJECT(cli));
+
+    return reply;
+}
+
+static void test_websock_bad_request(const void *opaque)
+{
+    const char *request = opaque;
+    g_autofree char *reply = test_websock_handshake_reply(request);
+
+    g_assert_true(g_str_has_prefix(reply, "HTTP/1.1 400 Bad Request\r\n"));
+}
+
+int main(int argc, char **argv)
+{
+    module_call_init(MODULE_INIT_QOM);
+    g_test_init(&argc, &argv, NULL);
+
+#define TEST_BAD_REQUEST(name, request)                         \
+    g_test_add_data_func("/io/channel/websock/bad-request/" name, \
+                         request, test_websock_bad_request)
+
+    /*
+     * A greeting with no space at all used to leave the response buffer
+     * empty, which drove the handshake into a zero length write.
+     */
+    TEST_BAD_REQUEST("no-space", "stats\r\nx\r\n\r\n");
+    TEST_BAD_REQUEST("method-only", "GET\r\nx\r\n\r\n");
+    TEST_BAD_REQUEST("no-version", "GET /\r\nx\r\n\r\n");
+    TEST_BAD_REQUEST("bad-method", "POST / HTTP/1.1\r\nx: y\r\n\r\n");
+    TEST_BAD_REQUEST("bad-version", "GET / HTTP/1.0\r\nx: y\r\n\r\n");
+
+    return g_test_run();
+}
-- 
2.55.0

[PULL v2 08/13] io/channel-websock: do not lose QIO_CHANNEL_ERR_BLOCK while reading

[Daniel P. Berrangé]

From: Denis V. Lunev <den@openvz.org>

qio_channel_websock_handshake_read() folds every negative return from
qio_channel_read() into -1. QIO_CHANNEL_ERR_BLOCK leaves errp unset, so
qio_channel_websock_handshake_io() then hands a NULL Error to
error_get_pretty() and QEMU dies.

The master channel is non-blocking and, for a wss:// client, is a TLS
channel. A G_IO_IN wakeup carrying only part of a TLS record makes
gnutls report EAGAIN, which is all it takes to reach this before the
client has authenticated.

ERR_BLOCK here means the headers are not complete yet, which is what a
0 return already tells the caller. Report it that way and keep waiting.
The watch is level triggered, so an incomplete record sitting in the
socket spins the main loop until the rest of it arrives. That is
bounded by the round trip and is what every reader layered over TLS
already does.

Fixes: 2d1d0e70cf3e ("io: add QIOChannelWebsock class")
Fixes: CVE-2026-84788
Cc: qemu-stable@nongnu.org
Cc: Daniel P. Berrangé <berrange@redhat.com>
Cc: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Denis V. Lunev <den@openvz.org>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
---
 io/channel-websock.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/io/channel-websock.c b/io/channel-websock.c
index 8f27b1f12b..461abcae48 100644
--- a/io/channel-websock.c
+++ b/io/channel-websock.c
@@ -492,6 +492,9 @@ static int qio_channel_websock_handshake_read(QIOChannelWebsock *ioc,
     buffer_reserve(&ioc->encinput, want);
     ret = qio_channel_read(ioc->master,
                            (char *)buffer_end(&ioc->encinput), want, errp);
+    if (ret == QIO_CHANNEL_ERR_BLOCK) {
+        return 0;
+    }
     if (ret < 0) {
         return -1;
     }
-- 
2.55.0

[PULL v2 09/13] tests/unit: cover blocked IO during the websock handshake

[Daniel P. Berrangé]

From: Denis V. Lunev <den@openvz.org>

Add a channel which reports QIO_CHANNEL_ERR_BLOCK on demand, the way a
TLS channel does when a record arrives split across segments or when
the socket cannot take the whole reply at once, and drive the server
handshake through it in both directions. Without the fixes each
direction dereferences a NULL Error and the test dies on SIGSEGV.

Cc: Daniel P. Berrangé <berrange@redhat.com>
Cc: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Denis V. Lunev <den@openvz.org>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
---
 tests/unit/test-io-channel-websock.c | 150 ++++++++++++++++++++++++++-
 1 file changed, 147 insertions(+), 3 deletions(-)

diff --git a/tests/unit/test-io-channel-websock.c b/tests/unit/test-io-channel-websock.c
index 2a55a4bcdf..88da24f993 100644
--- a/tests/unit/test-io-channel-websock.c
+++ b/tests/unit/test-io-channel-websock.c
@@ -12,6 +12,123 @@
 #include "qapi/error.h"
 #include "qemu/module.h"
 #include "qemu/sockets.h"
+#include "qom/object.h"
+
+#define TYPE_QIO_CHANNEL_STALL "qio-channel-stall"
+OBJECT_DECLARE_SIMPLE_TYPE(QIOChannelStall, QIO_CHANNEL_STALL)
+
+/*
+ * Reports QIO_CHANNEL_ERR_BLOCK for the first @rstalls reads and @wstalls
+ * writes, the way a TLS channel does when a record arrives split across TCP
+ * segments or the socket cannot take the whole reply at once.
+ */
+struct QIOChannelStall {
+    QIOChannel parent;
+    QIOChannel *master;
+    unsigned rstalls;
+    unsigned wstalls;
+};
+
+static ssize_t qio_channel_stall_readv(QIOChannel *ioc,
+                                       const struct iovec *iov,
+                                       size_t niov,
+                                       int **fds,
+                                       size_t *nfds,
+                                       int flags,
+                                       Error **errp)
+{
+    QIOChannelStall *sioc = QIO_CHANNEL_STALL(ioc);
+
+    if (sioc->rstalls) {
+        sioc->rstalls--;
+        return QIO_CHANNEL_ERR_BLOCK;
+    }
+    return qio_channel_readv_full(sioc->master, iov, niov, fds, nfds,
+                                  flags, errp);
+}
+
+static ssize_t qio_channel_stall_writev(QIOChannel *ioc,
+                                        const struct iovec *iov,
+                                        size_t niov,
+                                        int *fds,
+                                        size_t nfds,
+                                        int flags,
+                                        Error **errp)
+{
+    QIOChannelStall *sioc = QIO_CHANNEL_STALL(ioc);
+
+    if (sioc->wstalls) {
+        sioc->wstalls--;
+        return QIO_CHANNEL_ERR_BLOCK;
+    }
+    return qio_channel_writev_full(sioc->master, iov, niov, fds, nfds,
+                                   flags, errp);
+}
+
+static int qio_channel_stall_set_blocking(QIOChannel *ioc, bool enabled,
+                                          Error **errp)
+{
+    QIOChannelStall *sioc = QIO_CHANNEL_STALL(ioc);
+
+    return qio_channel_set_blocking(sioc->master, enabled, errp) ? 0 : -1;
+}
+
+static int qio_channel_stall_close(QIOChannel *ioc, Error **errp)
+{
+    QIOChannelStall *sioc = QIO_CHANNEL_STALL(ioc);
+
+    return qio_channel_close(sioc->master, errp);
+}
+
+static GSource *qio_channel_stall_create_watch(QIOChannel *ioc,
+                                               GIOCondition condition)
+{
+    QIOChannelStall *sioc = QIO_CHANNEL_STALL(ioc);
+
+    return qio_channel_create_watch(sioc->master, condition);
+}
+
+static void qio_channel_stall_finalize(Object *obj)
+{
+    QIOChannelStall *sioc = QIO_CHANNEL_STALL(obj);
+
+    object_unref(OBJECT(sioc->master));
+}
+
+static void qio_channel_stall_class_init(ObjectClass *klass,
+                                         const void *class_data G_GNUC_UNUSED)
+{
+    QIOChannelClass *ioc_klass = QIO_CHANNEL_CLASS(klass);
+
+    ioc_klass->io_writev = qio_channel_stall_writev;
+    ioc_klass->io_readv = qio_channel_stall_readv;
+    ioc_klass->io_set_blocking = qio_channel_stall_set_blocking;
+    ioc_klass->io_close = qio_channel_stall_close;
+    ioc_klass->io_create_watch = qio_channel_stall_create_watch;
+}
+
+static const TypeInfo qio_channel_stall_info = {
+    .parent = TYPE_QIO_CHANNEL,
+    .name = TYPE_QIO_CHANNEL_STALL,
+    .instance_size = sizeof(QIOChannelStall),
+    .instance_finalize = qio_channel_stall_finalize,
+    .class_init = qio_channel_stall_class_init,
+};
+
+static QIOChannelStall *qio_channel_stall_new(QIOChannel *master,
+                                              unsigned rstalls,
+                                              unsigned wstalls)
+{
+    QIOChannelStall *sioc = QIO_CHANNEL_STALL(
+        object_new(TYPE_QIO_CHANNEL_STALL));
+
+    object_ref(OBJECT(master));
+    sioc->master = master;
+    sioc->rstalls = rstalls;
+    sioc->wstalls = wstalls;
+
+    return sioc;
+}
 
 typedef struct {
     bool finished;
@@ -31,10 +148,12 @@ static void test_websock_handshake_done(QIOTask *task, gpointer opaque)
  * the server wrote back, NUL terminated. The handshake is expected to
  * fail; the point of the test is the HTTP response that goes with it.
  */
-static char *test_websock_handshake_reply(const char *request)
+static char *test_websock_handshake_reply(const char *request,
+                                          unsigned rstalls, unsigned wstalls)
 {
     QIOChannelWebsockHandshake res = { false, false };
     QIOChannelSocket *cli, *srv;
+    QIOChannelStall *stall;
     QIOChannelWebsock *wioc;
     GMainContext *mainloop;
     int channel[2];
@@ -48,7 +167,8 @@ static char *test_websock_handshake_reply(const char *request)
     qio_channel_set_blocking(QIO_CHANNEL(srv), false, &error_abort);
     qio_channel_set_blocking(QIO_CHANNEL(cli), false, &error_abort);
 
-    wioc = qio_channel_websock_new_server(QIO_CHANNEL(srv));
+    stall = qio_channel_stall_new(QIO_CHANNEL(srv), rstalls, wstalls);
+    wioc = qio_channel_websock_new_server(QIO_CHANNEL(stall));
     qio_channel_websock_handshake(wioc, test_websock_handshake_done,
                                   &res, NULL);
 
@@ -68,6 +188,7 @@ static char *test_websock_handshake_reply(const char *request)
     }
 
     object_unref(OBJECT(wioc));
+    object_unref(OBJECT(stall));
     object_unref(OBJECT(srv));
     object_unref(OBJECT(cli));
 
@@ -77,7 +198,23 @@ static char *test_websock_handshake_reply(const char *request)
 static void test_websock_bad_request(const void *opaque)
 {
     const char *request = opaque;
-    g_autofree char *reply = test_websock_handshake_reply(request);
+    g_autofree char *reply = test_websock_handshake_reply(request, 0, 0);
+
+    g_assert_true(g_str_has_prefix(reply, "HTTP/1.1 400 Bad Request\r\n"));
+}
+
+static void test_websock_stalled_read(const void *opaque)
+{
+    const char *request = opaque;
+    g_autofree char *reply = test_websock_handshake_reply(request, 1, 0);
+
+    g_assert_true(g_str_has_prefix(reply, "HTTP/1.1 400 Bad Request\r\n"));
+}
+
+static void test_websock_stalled_write(const void *opaque)
+{
+    const char *request = opaque;
+    g_autofree char *reply = test_websock_handshake_reply(request, 0, 1);
 
     g_assert_true(g_str_has_prefix(reply, "HTTP/1.1 400 Bad Request\r\n"));
 }
@@ -85,6 +222,7 @@ static void test_websock_bad_request(const void *opaque)
 int main(int argc, char **argv)
 {
     module_call_init(MODULE_INIT_QOM);
+    type_register_static(&qio_channel_stall_info);
     g_test_init(&argc, &argv, NULL);
 
 #define TEST_BAD_REQUEST(name, request)                         \
@@ -101,5 +239,11 @@ int main(int argc, char **argv)
     TEST_BAD_REQUEST("bad-method", "POST / HTTP/1.1\r\nx: y\r\n\r\n");
     TEST_BAD_REQUEST("bad-version", "GET / HTTP/1.0\r\nx: y\r\n\r\n");
 
+    /* A read which blocks before any header arrives is not a fatal error. */
+    g_test_add_data_func("/io/channel/websock/stalled-read",
+                         "stats\r\nx\r\n\r\n", test_websock_stalled_read);
+    g_test_add_data_func("/io/channel/websock/stalled-write",
+                         "stats\r\nx\r\n\r\n", test_websock_stalled_write);
+
     return g_test_run();
 }
-- 
2.55.0

[PULL v2 10/13] crypto: deprecate the AF_ALG crypto backend

[Daniel P. Berrangé]

Linux 7.2 has deprecated the AF_ALG crypto backend:

  https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=a67afb1884ba815079bd43d5c998e155e03b08b6

And has documented it to be always slower than userspace crypto:

  https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=5624ea54f3ba5c83d2e5503411a31a8be0278c1e

as a result of dropping support for zero-copy and hardware
accelerators:

  https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=7524070f26d8d347c26787dc297fb844baa26abf
  https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=ffdd2bc378953b525aca61902534e753f1f8e734

The main use case for the AF_ALG impl was to improve the performance
of virtio-crypto with the cryptodev-backend-builtin driver. In
practice this did not matter since 'cryptodev-backend-lkcf' can do
offload to the kernel via the keyctl syscall, and 'cryptodev-vhost-user'
can offload to an external process which can optionally integrate with
hardware accelerators without kernel assistance.

The AF_ALG backend has no user visible configuration options at runtime,
it is unconditionally tried with any use of the cipher APIs. So it does
not strictly have to go through the deprecation process, however, it is
left available initially in case there was an unexpected use case that
relies on it which may be faster with old kernels before the above Linux
commits.

Suggested-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@oss.qualcomm.com>
Reviewed-by: Cédric Le Goater <clg@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
---
 docs/about/deprecated.rst | 21 +++++++++++++++++++++
 meson.build               |  6 ++++++
 2 files changed, 27 insertions(+)

diff --git a/docs/about/deprecated.rst b/docs/about/deprecated.rst
index 05e4ce8cf1..98c32991c9 100644
--- a/docs/about/deprecated.rst
+++ b/docs/about/deprecated.rst
@@ -434,6 +434,27 @@ ABI is long-obsolete. We are therefore deprecating both OABI support
 and NWFPE emulation, and they will be removed in a future QEMU
 release.
 
+Build features
+--------------
+
+Crypto AF_ALG backend (since 11.2)
+----------------------------------
+
+The use of the AF_ALG backend for cryptography has been deprecated
+with no replacement.
+
+The AF_ALG interface is deprecated by Linux 7.2 and all support
+for hardware accelerators has been removed. It will thus always be
+slower than userspace crypto due to the overhead of copying data
+to kernel space. The GNUTLS, Nettle and GCrypt libraries supported
+by QEMU all include a variety of hardware optimized crypto
+implementations which should suffice for typical needs.
+
+For the virtio-crypto device, the 'cryptodev-backend-lkcf' backend
+can offload some operations to the kernel via the keyctl syscall,
+and the 'cryptodev-vhost-user' backend can offload the device
+backend to an external process which can integrate with crypto
+accelerators.
 
 Backwards compatibility
 -----------------------
diff --git a/meson.build b/meson.build
index 5c9de307b0..8528db72af 100644
--- a/meson.build
+++ b/meson.build
@@ -5070,3 +5070,9 @@ if not actually_reloc and (host_os == 'windows' or get_option('relocatable'))
   message('QEMU will have to be installed under ' + get_option('prefix') + '.')
   message('Use --disable-relocatable to remove this warning.')
 endif
+
+if get_option('crypto_afalg').enabled()
+    warning('Use of the AF_ALG crypto backend is deprecated, ' +
+            'since Linux 7.2 has deprecated the AF_ALG interface ' +
+            'and removed its ability to use hardware accelerators.')
+endif
-- 
2.55.0

[PULL v2 11/13] gitlab: use --emacs --quiet for checkpatch.pl instead of --terse

[Daniel P. Berrangé]

The default checkpatch.pl output includes a lot of "progress"
information...

   NN/MM Checking commit HASH (SUBJECT)
   total: 0 errors, 0 warnings, 105 lines checked

for large patch series, this results in alot of noise from
clean patches (which are the common case), obscuring the
info about the patch violations.

Since the alerts from gitlab job failures only include the
last few lines of log output, we want the checkpatch.pl
output to be highly relevant to the failure.

We previously addressed that by adding use of the --terse
flag in 7025111a199b97ae806817788bec50f456c47d85, but that
made the output a bit too terse. It no longer prints the
offending line of code, so understanding the CI job failure
now requires manually re-running checkpatch.pl locally.

Using the '--quiet' flag gets rid of the "Check commit HASH.."
messages which create noise, while adding '--emacs' causes
the WARNING/ERROR message lines to include the commit hash.

This is a more useful tradeoff for the CI job logs. The
output is limited to only patches which include code
violations, while still including the offending lines of
code in the output.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
---
 .gitlab-ci.d/check-patch.py | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/.gitlab-ci.d/check-patch.py b/.gitlab-ci.d/check-patch.py
index be13e6f77d..45be77295d 100755
--- a/.gitlab-ci.d/check-patch.py
+++ b/.gitlab-ci.d/check-patch.py
@@ -46,7 +46,11 @@
 
 print("\nChecking all commits since %s...\n" % ancestor, flush=True)
 
-ret = subprocess.run(["scripts/checkpatch.pl", "--terse", ancestor + "..."])
+# We don't want "noise" for clean patches, but do want to see
+# the full commit hash for each violation, along with the
+# offending patch content
+ret = subprocess.run(["scripts/checkpatch.pl", "--emacs", "--quiet",
+                      ancestor + "..."])
 
 if ret.returncode != 0:
     print("    ❌ FAIL one or more commits failed scripts/checkpatch.pl")
-- 
2.55.0

[PULL v2 12/13] configure: correctly honour --disable-containers

[Daniel P. Berrangé]

The configure script originally only probed for 'runc' command
in order to handle cross-compilation containers. The Makefile
under tests/docker would then probe again when used.

In order to eliminate the additional probeing, we changed
configure in commit c4ce04cfb7460d46c0262d437a7f35bb5c5fc449
to always probe for 'runc', so the result could be passed
along to tests/docker make rules.

That commit overlooked that the 'probe_target_compiler' func
was relying on '$runc' == 'no' as a proxy for the state of
the --disable-containers arg.

When we started unconditionally probing, that short cut no
longer works and we must explicitly check "$use_containers"
too.

Fixes: c4ce04cfb7460d46c0262d437a7f35bb5c5fc449
Reported-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Tested-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
---
 configure | 1 +
 1 file changed, 1 insertion(+)

diff --git a/configure b/configure
index 46cbcc085b..d3ad7d4062 100755
--- a/configure
+++ b/configure
@@ -1414,6 +1414,7 @@ probe_target_compiler() {
   esac
 
   for host in $container_hosts; do
+    test "$use_containers" = "yes" || continue
     test "$container_command" != "" || continue
     test "$host" = "$cpu" || continue
     case $target_arch in
-- 
2.55.0

Re: [PULL v2 12/13] configure: correctly honour --disable-containers

[Daniel P. Berrangé]

I'm dropping this patch from the PULL refresh, since Pierrick's
merged patches remove the container engine probing from configure,
so the bug this fixed appears to not exist any more.

On Fri, Sep 04, 2026 at 05:06:42PM +0100, Daniel P. Berrangé wrote:
> The configure script originally only probed for 'runc' command
> in order to handle cross-compilation containers. The Makefile
> under tests/docker would then probe again when used.
> 
> In order to eliminate the additional probeing, we changed
> configure in commit c4ce04cfb7460d46c0262d437a7f35bb5c5fc449
> to always probe for 'runc', so the result could be passed
> along to tests/docker make rules.
> 
> That commit overlooked that the 'probe_target_compiler' func
> was relying on '$runc' == 'no' as a proxy for the state of
> the --disable-containers arg.
> 
> When we started unconditionally probing, that short cut no
> longer works and we must explicitly check "$use_containers"
> too.
> 
> Fixes: c4ce04cfb7460d46c0262d437a7f35bb5c5fc449
> Reported-by: Alex Bennée <alex.bennee@linaro.org>
> Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
> Tested-by: Alex Bennée <alex.bennee@linaro.org>
> Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
> ---
>  configure | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/configure b/configure
> index 46cbcc085b..d3ad7d4062 100755
> --- a/configure
> +++ b/configure
> @@ -1414,6 +1414,7 @@ probe_target_compiler() {
>    esac
>  
>    for host in $container_hosts; do
> +    test "$use_containers" = "yes" || continue
>      test "$container_command" != "" || continue
>      test "$host" = "$cpu" || continue
>      case $target_arch in
> -- 
> 2.55.0
> 

With regards,
Daniel
-- 
|: https://berrange.com       ~~        https://hachyderm.io/@berrange :|
|: https://libvirt.org          ~~          https://entangle-photo.org :|
|: https://pixelfed.art/berrange   ~~    https://fstop138.berrange.com :|

[PULL v2 13/13] docs/system/security: exclude uninitialized stack variables as bugs

[Daniel P. Berrangé]

The -ftrivial-auto-var-init=zero usage guarantees implicit zero
initializers for all stack variables. Thus most bug reports relying
on undefined behaviour from lack of variable initialization will
not be security issues, or even bugs.

Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
---
 docs/system/security.rst | 10 ++++++++++
 1 file changed, 10 insertions(+)

diff --git a/docs/system/security.rst b/docs/system/security.rst
index af626a4230..8c42d1a6d8 100644
--- a/docs/system/security.rst
+++ b/docs/system/security.rst
@@ -143,6 +143,16 @@ an issue as a normal bug.
   which case plain manipulation of the stream is not considered as
   an attack vector.
 
+* **uninitialized stack variables**. If the bug scenario relies on
+  undefined behaviour from stack variables that lack explicit
+  initialization, it will not usually be considered a security flaw.
+  The build system adds '-ftrivial-auto-var-init=zero', which is
+  available in both the supported compilers (GCC and CLang) and
+  ensures all stack variables have implicit zero-initializers.
+  This eliminates undefined behaviour and usually gives the
+  correct desired initialization value, eliminating most of the
+  bug scenarios wrt uninitialized stack variables.
+
 * **low severity impact**. As a catch all rule, issues which
   are judged to have a "low" severity impact on the system will
   usually not justify handling as security bugs, nor assignment
-- 
2.55.0

Re: [PULL v2 00/13] Misc fixes patches

[Peter Maydell]

On Fri, 4 Sept 2026 at 17:07, Daniel P. Berrangé <berrange@redhat.com> wrote:
>
> The following changes since commit 99e54ab5e7a6efc945af6d5661842155d1f3fc7a:
>
>   Merge tag 'hw-misc-20260903' of https://github.com/philmd/qemu into staging (2026-09-03 16:43:24 +0100)
>
> are available in the Git repository at:
>
>   https://gitlab.com/berrange/qemu tags/misc-fixes-pull-request
>
> for you to fetch changes up to 39dfe7cf8e87d6de975bae6c10c3712f4df56ee6:
>
>   docs/system/security: exclude uninitialized stack variables as bugs (2026-09-04 14:29:16 +0100)
>
> ----------------------------------------------------------------
> Merge crypto, I/O and misc fixes
>
>  * Deprecated the AF_ALG crypto backend
>  * Improve checkpatch output in CI jobs
>  * Fix --disable-containers arg handling in configure
>  * Document security policy for uninitialized stack variables
>  * Fix multiple denial of service flaws in websockets
>  * Fix error handling in some x509 APIs
>

Hi -- I'm afraid this has a conflict in configure: could you rebase
and fix, please?

thanks
-- PMM