mdadm: Create array with sync del gendisk mode

kernel patch 9e59d609763f ('md: call del_gendisk in control path') calls
del_gendisk in sync way. After the patch mentioned just now, device node
(/dev/md0 .e.g) will disappear after mdadm --stop command. It resolves the
problem raid can be created again because raid can be created when opening
device node. Then regression tests will be interrupted.

But it causes an error when assembling array which has been fixed by pr182.
So people can't assemble array if they use new kernel and old mdadm. So
in kernel space, 25db5f284fb8 ('md: add legacy_async_del_gendisk mod') is
used to fix this problem. The default is async mode.

async del mode will be removed in future. We'll start use sync del mode in
new mdadm version. So people will not see failure when upgrading to the
new mdadm version with sync del mode.

Signed-off-by: Xiao Ni <xni@redhat.com>

diff --git a/mdadm.h b/mdadm.h
index 84bd2c915fc25a62f70fd5c838b667a288cd6576..7dcb20ed1f34d5b4a13f2017890c2bb7afcdbd03 100644 (file)
--- a/mdadm.h
+++ b/mdadm.h
@@ -141,6 +141,8 @@ struct dlm_lksb {
 #define MDMON_DIR "/run/mdadm"
 #endif /* MDMON_DIR */
 
+#define MD_MOD_ASYNC_DEL_GENDISK "legacy_async_del_gendisk"
+
 /* FAILED_SLOTS is where to save files storing recent removal of array
  * member in order to allow future reuse of disk inserted in the same
  * slot for array recovery
@@ -855,6 +857,7 @@ extern int restore_stripes(int *dest, unsigned long long *offsets,
                           unsigned long long start, unsigned long long length,
                           char *src_buf);
 extern bool sysfs_is_libata_allow_tpm_enabled(const int verbose);
+extern bool init_md_mod_param(void);
 
 #ifndef Sendmail
 #define Sendmail "/usr/lib/sendmail -t"

diff --git a/mdopen.c b/mdopen.c
index 57252b646137ce0616ee2c24ad134e36f15d8e5a..b685603d6de546a21060a81b941b23243ea8b71a 100644 (file)
--- a/mdopen.c
+++ b/mdopen.c
@@ -148,6 +148,11 @@ int create_mddev(char *dev, char *name, int trustworthy,
        char devnm[32];
        char cbuf[400];
 
+       if (!init_md_mod_param()) {
+               pr_err("init md module parameters fail\n");
+               return -1;
+       }
+
        if (!udev_is_available())
                block_udev = 0;
 

diff --git a/util.c b/util.c
index 5d6fe800d6669d4404c1a41caa3ca62b6f9257f4..146f38fddd82ffaa14b1942c7404244e27cc3100 100644 (file)
--- a/util.c
+++ b/util.c
@@ -2559,3 +2559,48 @@ bool is_file(const char *path)
 
        return true;
 }
+
+bool set_md_mod_parameter(const char *name, const char *value)
+{
+       char path[256];
+       int fd;
+       bool ret = true;
+
+       snprintf(path, sizeof(path), "/sys/module/md_mod/parameters/%s", name);
+
+       fd = open(path, O_WRONLY);
+       if (fd < 0) {
+               pr_err("Can't open %s\n", path);
+               return false;
+       }
+
+       if (write(fd, value, strlen(value)) != (ssize_t)strlen(value)) {
+               pr_err("Failed to write to %s\n", path);
+               ret = false;
+       }
+
+       close(fd);
+       return ret;
+}
+
+/* Init kernel md_mod parameters here if needed */
+bool init_md_mod_param(void)
+{
+       bool ret = true;
+
+       /*
+        * In kernel 9e59d609763f calls del_gendisk in sync way. So device
+        * node can be removed after stop command. But it can introduce a
+        * regression which can be fixed by github pr182. New mdadm version
+        * with pr182 can work well with new kernel. But users who don't
+        * update mdadm and update to new kernel, they can't assemble array
+        * anymore. So kernel adds a kernel parameter legacy_async_del_gendisk
+        * and uses async as default.
+        * We'll use sync mode since 6.18 rather than async mode. So in future
+        * the kernel parameter will be removed.
+        */
+       if (get_linux_version() >= 6018000)
+               ret = set_md_mod_parameter(MD_MOD_ASYNC_DEL_GENDISK, "N");
+
+       return ret;
+}