Linux gained new close_range() syscall (in v5.9) that allows
closing a range of FDs in a single syscall. Ideally, we would use
it to close FDs when spawning a process (e.g. via virCommand
module).
Glibc has close_range() wrapper over the syscall, which falls
back to iterative closing of all FDs inside the range if running
under older kernel. We don't wan that as in that case we might
just close opened FDs (see Linux version of
virCommandMassClose()). And musl doesn't have close_range() at
all. Therefore, call syscall directly.
Now, mass close of FDs happen in a fork()-ed off child. While it
could detect whether the kernel does support close_range(), it
has no way of passing this info back to the parent and thus each
child would need to query it again and again.
Since this can't change while we are running we can cache the
information - hence virCloseRangeInit().
Signed-off-by: Michal Privoznik <mprivozn(a)redhat.com>
---
src/libvirt_private.syms | 3 ++
src/util/virfile.c | 89 ++++++++++++++++++++++++++++++++++++++++
src/util/virfile.h | 4 ++
3 files changed, 96 insertions(+)
diff --git a/src/libvirt_private.syms b/src/libvirt_private.syms
index da60c965dd..3782f7f3c7 100644
--- a/src/libvirt_private.syms
+++ b/src/libvirt_private.syms
@@ -2269,6 +2269,9 @@ saferead;
safewrite;
safezero;
virBuildPathInternal;
+virCloseRange;
+virCloseRangeInit;
+virCloseRangeIsSupported;
virDirClose;
virDirCreate;
virDirIsEmpty;
diff --git a/src/util/virfile.c b/src/util/virfile.c
index fe456596ae..7696910e00 100644
--- a/src/util/virfile.c
+++ b/src/util/virfile.c
@@ -87,6 +87,7 @@
#include "virlog.h"
#include "virprocess.h"
#include "virstring.h"
+#include "virthread.h"
#include "virutil.h"
#include "virsocket.h"
@@ -109,6 +110,9 @@ VIR_LOG_INIT("util.file");
# define O_DIRECT 0
#endif
+static virOnceControl virCloseRangeOnce = VIR_ONCE_CONTROL_INITIALIZER;
+static bool virCloseRangeSupported;
+
int virFileClose(int *fdptr, virFileCloseFlags flags)
{
int saved_errno = 0;
@@ -176,6 +180,91 @@ FILE *virFileFdopen(int *fdptr, const char *mode)
}
+static int
+virCloseRangeImpl(unsigned int first G_GNUC_UNUSED,
+ unsigned int last G_GNUC_UNUSED)
+{
+#if defined(WITH_SYS_SYSCALL_H) && defined(__NR_close_range)
+ return syscall(__NR_close_range, first, last, 0);
+#endif
+
+ errno = ENOSYS;
+ return -1;
+}
+
+
+static void
+virCloseRangeOnceInit(void)
+{
+ int fd[2] = { -1, -1};
+
+ if (virPipeQuiet(fd) < 0)
+ return;
+
+ VIR_FORCE_CLOSE(fd[1]);
+ if (virCloseRangeImpl(fd[0], fd[0]) < 0) {
+ VIR_FORCE_CLOSE(fd[0]);
+ return;
+ }
+
+ virCloseRangeSupported = true;
+}
+
+
+/**
+ * virCloseRange:
+ *
+ * Closes all open file descriptors from @first to @last (included).
+ *
+ * Returns: 0 on success,
+ * -1 on failure (with errno set).
+ */
+int
+virCloseRange(unsigned int first,
+ unsigned int last)
+{
+ if (virCloseRangeInit() < 0)
+ return -1;
+
+ if (!virCloseRangeSupported) {
+ errno = ENOSYS;
+ return -1;
+ }
+
+ return virCloseRangeImpl(first, last);
+}
+
+
+/**
+ * virCloseRangeInit:
+ *
+ * Detects whether close_range() is available and cache the result.
+ */
+int
+virCloseRangeInit(void)
+{
+ if (virOnce(&virCloseRangeOnce, virCloseRangeOnceInit) < 0)
+ return -1;
+
+ return 0;
+}
+
+
+/**
+ * virCloseRangeIsSupported:
+ *
+ * Returns whether close_range() is supported or not.
+ */
+bool
+virCloseRangeIsSupported(void)
+{
+ if (virCloseRangeInit() < 0)
+ return false;
+
+ return virCloseRangeSupported;
+}
+
+
/**
* virFileDirectFdFlag:
*
diff --git a/src/util/virfile.h b/src/util/virfile.h
index 60bb1d64e7..be0b02fdf0 100644
--- a/src/util/virfile.h
+++ b/src/util/virfile.h
@@ -61,6 +61,10 @@ static inline void virForceCloseHelper(int *fd)
ignore_value(virFileClose(fd, VIR_FILE_CLOSE_PRESERVE_ERRNO));
}
+int virCloseRange(unsigned int from, unsigned int to);
+int virCloseRangeInit(void);
+bool virCloseRangeIsSupported(void);
+
/* For use on normal paths; caller must check return value,
and failure sets errno per close. */
#define VIR_CLOSE(FD) virFileClose(&(FD), 0)
--
2.41.0