[PATCH 1/2] btrfs-progs: scan: cleanup, return errno when we have one

[PATCH 1/2] btrfs-progs: scan: cleanup, return errno when we have one

[Anand Jain]

Functions called in cmd_device_scan() and its return codes as below:

btrfs_forget_devices() returns -errno pass down from ioctl or -errno of
open().
btrfs_register_one_device() returns -errno pass down from the kernel
ioctl or -errno from open().
btrfs_scan_devices() (after the patch) returns error code from libblk who's
return error code matches to -errno or 0.
btrfs_register_all_devices() returns number of devices that failed to
regester (as of now) which no one uses. However if
btrfs_register_all_devices() returns error, we error_on() and return 0.

So we can safely return the error code instead of return !!ret;

Signed-off-by: Anand Jain <anand.jain@oracle.com>
---
 cmds-device.c | 12 +++++++-----
 1 file changed, 7 insertions(+), 5 deletions(-)

diff --git a/cmds-device.c b/cmds-device.c
index e3e30b6d5ded..bb3dc78149f8 100644
--- a/cmds-device.c
+++ b/cmds-device.c
@@ -359,13 +359,13 @@ static int cmd_device_scan(int argc, char **argv)
 
 		if (is_block_device(argv[i]) != 1) {
 			error("not a block device: %s", argv[i]);
-			ret = 1;
+			ret = EINVAL;
 			goto out;
 		}
 		path = canonicalize_path(argv[i]);
 		if (!path) {
 			error("could not canonicalize path '%s': %m", argv[i]);
-			ret = 1;
+			ret = EINVAL;
 			goto out;
 		}
 		if (forget) {
@@ -376,8 +376,8 @@ static int cmd_device_scan(int argc, char **argv)
 			}
 		} else {
 			printf("Scanning for btrfs filesystems on '%s'\n", path);
-			if (btrfs_register_one_device(path) != 0) {
-				ret = 1;
+			ret = btrfs_register_one_device(path);
+			if (ret < 0) {
 				free(path);
 				goto out;
 			}
@@ -386,7 +386,9 @@ static int cmd_device_scan(int argc, char **argv)
 	}
 
 out:
-	return !!ret;
+	if (ret < 0)
+		return -ret;
+	return ret;
 }
 
 static const char * const cmd_device_ready_usage[] = {
-- 
1.8.3.1

[PATCH 2/2] btrfs-progs: scan: pass blkid_get_cache error code

[Anand Jain]

blkid_get_cache() returns error code which is -errno. So we can use them
directly.

Signed-off-by: Anand Jain <anand.jain@oracle.com>
---
Ref:
blkid_get_cache() code:
https://github.com/karelzak/util-linux/blob/master/libblkid/src/cache.c#L93
https://github.com/karelzak/util-linux/blob/master/libblkid/src/blkidP.h#L307
 utils.c | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)

diff --git a/utils.c b/utils.c
index 9e26c884cc6c..c6cdc8f01dc1 100644
--- a/utils.c
+++ b/utils.c
@@ -1994,9 +1994,11 @@ int btrfs_scan_devices(void)
 	if (btrfs_scan_done)
 		return 0;
 
-	if (blkid_get_cache(&cache, NULL) < 0) {
-		error("blkid cache get failed");
-		return 1;
+	ret = blkid_get_cache(&cache, NULL);
+	if (ret < 0) {
+		errno = -ret;
+		error("blkid cache get failed: %m");
+		return ret;
 	}
 	blkid_probe_all(cache);
 	iter = blkid_dev_iterate_begin(cache);
-- 
1.8.3.1

Re: [PATCH 1/2] btrfs-progs: scan: cleanup, return errno when we have one

[David Sterba]

On Fri, Mar 29, 2019 at 01:49:53PM +0800, Anand Jain wrote:
> @@ -386,7 +386,9 @@ static int cmd_device_scan(int argc, char **argv)
>  	}
>  
>  out:
> -	return !!ret;
> +	if (ret < 0)
> +		return -ret;
> +	return ret;

No, cmd_device_scan as the command handler returns the error code that's
intepreted by shell. Most commands do just 0 and 1, in documented case
there are other values, but we can't return errno here.

Re: [PATCH 2/2] btrfs-progs: scan: pass blkid_get_cache error code

[David Sterba]

On Fri, Mar 29, 2019 at 01:49:54PM +0800, Anand Jain wrote:
> blkid_get_cache() returns error code which is -errno. So we can use them
> directly.
> 
> Signed-off-by: Anand Jain <anand.jain@oracle.com>
> ---
> Ref:
> blkid_get_cache() code:
> https://github.com/karelzak/util-linux/blob/master/libblkid/src/cache.c#L93
> https://github.com/karelzak/util-linux/blob/master/libblkid/src/blkidP.h#L307

This is internal header of blkid and incidentally the error numbers
match errnos, but I don't think we should rely on that. Does blkid have
a function that translates the code to string, similar to strerror?

Re: [PATCH 1/2] btrfs-progs: scan: cleanup, return errno when we have one

[Anand Jain]

On 15/5/19 10:26 PM, David Sterba wrote:
> On Fri, Mar 29, 2019 at 01:49:53PM +0800, Anand Jain wrote:
>> @@ -386,7 +386,9 @@ static int cmd_device_scan(int argc, char **argv)
>>   	}
>>   
>>   out:
>> -	return !!ret;
>> +	if (ret < 0)
>> +		return -ret;
>> +	return ret;
> 
> No, cmd_device_scan as the command handler returns the error code that's
> intepreted by shell. Most commands do just 0 and 1, in documented case
> there are other values, but we can't return errno here.
> 

Ok. errno is better error informative though. Interpreted by shell seems 
to be a roadblock.

Re: [PATCH 2/2] btrfs-progs: scan: pass blkid_get_cache error code

[Anand Jain]

On 15/5/19 10:29 PM, David Sterba wrote:
> On Fri, Mar 29, 2019 at 01:49:54PM +0800, Anand Jain wrote:
>> blkid_get_cache() returns error code which is -errno. So we can use them
>> directly.
>>
>> Signed-off-by: Anand Jain <anand.jain@oracle.com>
>> ---
>> Ref:
>> blkid_get_cache() code:
>> https://github.com/karelzak/util-linux/blob/master/libblkid/src/cache.c#L93
>> https://github.com/karelzak/util-linux/blob/master/libblkid/src/blkidP.h#L307
> 
> This is internal header of blkid and incidentally the error numbers
> match errnos, but I don't think we should rely on that. Does blkid have
> a function that translates the code to string, similar to strerror?
> 

Uh. No there isn't BLKID_ERR.. to string conversion.