Clang's version of -Wunused-but-set-variable recently gained support for
unary operations, which reveals two unused variables:

  fs/btrfs/block-group.c:2949:6: error: variable 'num_started' set but not used [-Werror,-Wunused-but-set-variable]
          int num_started = 0;
              ^
  fs/btrfs/block-group.c:3116:6: error: variable 'num_started' set but not used [-Werror,-Wunused-but-set-variable]
          int num_started = 0;
              ^
  2 errors generated.

These variables appear to be unused from their introduction, so just
remove them to silence the warnings.

Cc: stable@vger.kernel.org
Fixes: c9dc4c657850 ("Btrfs: two stage dirty block group writeout")
Fixes: 1bbc621ef284 ("Btrfs: allow block group cache writeout outside critical section in commit")
Link: https://github.com/ClangBuiltLinux/linux/issues/1614
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
---

I am requesting a stable backport because this is visible with
allmodconfig, which enables CONFIG_WERROR, breaking the build.

To quote Linus:

"EVERYBODY should have CONFIG_WERROR=y on at least x86-64 and other
serious architectures, unless you have some completely random
experimental (and broken) compiler."

https://lore.kernel.org/r/CAHk-=wifoM9VOp-55OZCRcO9MnqQ109UTuCiXeZ-eyX_JcNVGg@mail.gmail.com/

 fs/btrfs/block-group.c | 4 ----
 1 file changed, 4 deletions(-)

diff --git a/fs/btrfs/block-group.c b/fs/btrfs/block-group.c
index c22d287e020b..9ad265066225 100644
--- a/fs/btrfs/block-group.c
+++ b/fs/btrfs/block-group.c
@@ -2946,7 +2946,6 @@ int btrfs_start_dirty_block_groups(struct btrfs_trans_handle *trans)
 	struct btrfs_path *path = NULL;
 	LIST_HEAD(dirty);
 	struct list_head *io = &cur_trans->io_bgs;
-	int num_started = 0;
 	int loops = 0;
 
 	spin_lock(&cur_trans->dirty_bgs_lock);
@@ -3012,7 +3011,6 @@ int btrfs_start_dirty_block_groups(struct btrfs_trans_handle *trans)
 			cache->io_ctl.inode = NULL;
 			ret = btrfs_write_out_cache(trans, cache, path);
 			if (ret == 0 && cache->io_ctl.inode) {
-				num_started++;
 				should_put = 0;
 
 				/*
@@ -3113,7 +3111,6 @@ int btrfs_write_dirty_block_groups(struct btrfs_trans_handle *trans)
 	int should_put;
 	struct btrfs_path *path;
 	struct list_head *io = &cur_trans->io_bgs;
-	int num_started = 0;
 
 	path = btrfs_alloc_path();
 	if (!path)
@@ -3171,7 +3168,6 @@ int btrfs_write_dirty_block_groups(struct btrfs_trans_handle *trans)
 			cache->io_ctl.inode = NULL;
 			ret = btrfs_write_out_cache(trans, cache, path);
 			if (ret == 0 && cache->io_ctl.inode) {
-				num_started++;
 				should_put = 0;
 				list_add_tail(&cache->io_list, io);
 			} else {

base-commit: d3e29967079c522ce1c5cab0e9fab2c280b977eb
-- 
2.35.1

On Thu, Mar 24, 2022 at 08:36:45AM -0700, Nathan Chancellor wrote:
> Clang's version of -Wunused-but-set-variable recently gained support for
> unary operations, which reveals two unused variables:
> 
>   fs/btrfs/block-group.c:2949:6: error: variable 'num_started' set but not used [-Werror,-Wunused-but-set-variable]
>           int num_started = 0;
>               ^
>   fs/btrfs/block-group.c:3116:6: error: variable 'num_started' set but not used [-Werror,-Wunused-but-set-variable]
>           int num_started = 0;
>               ^
>   2 errors generated.
> 
> These variables appear to be unused from their introduction, so just
> remove them to silence the warnings.
> 
> Cc: stable@vger.kernel.org
> Fixes: c9dc4c657850 ("Btrfs: two stage dirty block group writeout")
> Fixes: 1bbc621ef284 ("Btrfs: allow block group cache writeout outside critical section in commit")
> Link: https://github.com/ClangBuiltLinux/linux/issues/1614
> Signed-off-by: Nathan Chancellor <nathan@kernel.org>

Added to misc-next, thanks.

> I am requesting a stable backport because this is visible with
> allmodconfig, which enables CONFIG_WERROR, breaking the build.

Yeah warning fixes are accepted in stable trees and we care about
warning-free builds. My gcc 11.1 does not produce any warning with
-Wunused-but-set-variable and we have extended the set of warnings in
the directory fs/btrfs/ that also includes that one.

> To quote Linus:
> 
> "EVERYBODY should have CONFIG_WERROR=y on at least x86-64 and other
> serious architectures, unless you have some completely random
> experimental (and broken) compiler."

I have CONFIG_WERROR disabled, for own development it's quite annoying
when build fails outside of the code I care about, while I apply visual
checks of warning-free builds.