From 6c5f4650f23dba2188ceba75a36f40bbf895216e Mon Sep 17 00:00:00 2001
Message-Id: <6c5f4650f23dba2188ceba75a36f40bbf895216e.1386932212.git.jdenemar@redhat.com>
From: Michal Privoznik <mprivozn@redhat.com>
Date: Tue, 10 Dec 2013 15:42:40 +0100
Subject: [PATCH] qemu_process: Read errors from child

https://bugzilla.redhat.com/show_bug.cgi?id=1035955

There's a window when starting a qemu process between fork() and exec()
during which we are doing things that may fail but not tunnelling the
error to the daemon. This is basically all within qemuProcessHook().
So whenever we fail in something, e.g. placing a process onto numa node,
users are left with:

    error: Child quit during startup handshake: Input/output error

while the original error is thrown into the domain log:

    libvirt:  error : internal error: NUMA memory tuning in 'preferred'
    mode only supports single node

Hence, we should read the log file and search for the error message and
report it to users.

Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
(cherry picked from commit 11e44d66fd93a8614332e46977cafc1d5fbebe7a)
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
---
 src/qemu/qemu_process.c | 66 +++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 66 insertions(+)

diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c
index 40e3754..a4b757c 100644
--- a/src/qemu/qemu_process.c
+++ b/src/qemu/qemu_process.c
@@ -1543,6 +1543,70 @@ cleanup:
 
 
 /*
+ * Read domain log and probably overwrite error if there's one in
+ * the domain log file. This function exists to cover the small
+ * window between fork() and exec() during which child may fail
+ * by libvirt's hand, e.g. placing onto a NUMA node failed.
+ */
+static int
+qemuProcessReadChildErrors(virQEMUDriverPtr driver,
+                           virDomainObjPtr vm,
+                           off_t originalOff)
+{
+    int ret = -1;
+    int logfd;
+    off_t off = 0;
+    ssize_t bytes;
+    char buf[1024] = {0};
+    char *eol, *filter_next = buf;
+
+    if ((logfd = qemuDomainOpenLog(driver, vm, originalOff)) < 0)
+        goto cleanup;
+
+    while (off < sizeof(buf) - 1) {
+        bytes = saferead(logfd, buf + off, sizeof(buf) - off - 1);
+        if (bytes < 0) {
+            VIR_WARN("unable to read from log file: %s",
+                     virStrerror(errno, buf, sizeof(buf)));
+            goto cleanup;
+        }
+
+        off += bytes;
+        buf[off] = '\0';
+
+        if (bytes == 0)
+            break;
+
+        while ((eol = strchr(filter_next, '\n'))) {
+            *eol = '\0';
+            if (STRPREFIX(filter_next, "libvirt: ")) {
+                filter_next = eol + 1;
+                *eol = '\n';
+                break;
+            } else {
+                memmove(filter_next, eol + 1, off - (eol - buf));
+                off -= eol + 1 - filter_next;
+            }
+        }
+    }
+
+    if (off > 0) {
+        /* Found an error in the log. Report it */
+        virResetLastError();
+        virReportError(VIR_ERR_INTERNAL_ERROR,
+                       _("Process exited prior to exec: %s"),
+                       buf);
+    }
+
+    ret = 0;
+
+cleanup:
+    VIR_FORCE_CLOSE(logfd);
+    return ret;
+}
+
+
+/*
  * Look at a chunk of data from the QEMU stdout logs and try to
  * find a TTY device, as indicated by a line like
  *
@@ -3877,6 +3941,8 @@ int qemuProcessStart(virConnectPtr conn,
 
     VIR_DEBUG("Waiting for handshake from child");
     if (virCommandHandshakeWait(cmd) < 0) {
+        /* Read errors from child that occurred between fork and exec. */
+        qemuProcessReadChildErrors(driver, vm, pos);
         goto cleanup;
     }
 
-- 
1.8.5.1