On 7/18/24 11:39, Jan Kara wrote:
> On Thu 18-07-24 00:14:24, Mirsad Todorovac wrote:
>>
>>
>> On 7/17/24 17:44, Jan Kara wrote:
>>> On Tue 16-07-24 19:17:05, Mirsad Todorovac wrote:
>>>> On 7/15/24 19:28, Jan Kara wrote:
>>>>> Hello Mirsad!
>>>>>
>>>>> On Wed 10-07-24 20:09:27, Mirsad Todorovac wrote:
>>>>>> On the linux-next vanilla next-20240709 tree, I have attempted the seed KCONFIG_SEED=0xEE7AB52F
>>>>>> which was known from before to trigger various errors in compile and build process.
>>>>>>
>>>>>> Though this might seem as contributing to channel noise, Linux refuses to build this config,
>>>>>> treating warnings as errors, using this build line:
>>>>>>
>>>>>> $ time nice make W=1 -k -j 36 |& tee ../err-next-20230709-01a.log; date
>>>>>>
>>>>>> As I know that the Chief Penguin doesn't like warnings, but I am also aware that there are plenty
>>>>>> left, there seems to be more tedious work ahead to make the compilers happy.
>>>>>>
>>>>>> The compiler output is:
>>>>>>
>>>>>> ---------------------------------------------------------------------------------------------------------
>>>>>> fs/reiserfs/do_balan.c: In function ‘balance_leaf_new_nodes_paste_whole’:
>>>>>> fs/reiserfs/do_balan.c:1147:13: error: variable ‘leaf_mi’ set but not used [-Werror=unused-but-set-variable]
>>>>>> 1147 | int leaf_mi;
>>>>>> | ^~~~~~~
>>>>>
>>>>> Frankly, I wouldn't bother with reiserfs. The warning is there for ages,
>>>>> the code is going to get removed in two releases, so I guess we can live
>>>>> with these warnings for a few more months...
>>>>
>>>> In essence I agree with you, but for sentimental reasons I would like to
>>>> keep it because it is my first journaling Linux system on Knoppix 🙂
>>>
>>> As much as I understand your sentiment (I have a bit of history with that
>>> fs as well) the maintenance cost isn't really worth it and most fs folks
>>> will celebrate when it's removed. We have already announced the removal
>>> year and half ago and I'm fully for executing that plan at the end of this
>>> year.
>>>
>>>> Patch is also simple and a no-brainer, as proposed by Mr. Cook:
>>>>
>>>> -------------------------------><------------------------------------------
>>>> diff --git a/fs/reiserfs/do_balan.c b/fs/reiserfs/do_balan.c
>>>> index 5129efc6f2e6..fbe73f267853 100644
>>>> --- a/fs/reiserfs/do_balan.c
>>>> +++ b/fs/reiserfs/do_balan.c
>>>> @@ -1144,7 +1144,9 @@ static void balance_leaf_new_nodes_paste_whole(struct tree_balance *tb,
>>>> {
>>>> struct buffer_head *tbS0 = PATH_PLAST_BUFFER(tb->tb_path);
>>>> int n = B_NR_ITEMS(tbS0);
>>>> +#ifdef CONFIG_REISERFS_CHECK
>>>> int leaf_mi;
>>>> +#endif
>>>
>>> Well, I would not like this even for actively maintained code ;) If you
>>> want to silence these warnings in this dead code, then I could live with
>>> something like:
>>>
>>> #if defined( CONFIG_REISERFS_CHECK )
>>> #define RFALSE(cond, format, args...) __RASSERT(!(cond), ....)
>>> #else
>>> - #define RFALSE( cond, format, args... ) do {;} while( 0 )
>>> + #define RFALSE( cond, format, args... ) do { (void)cond; } while( 0 )
>>> #endif
>>
>> Yes, one line change is much smarter than 107 line patch of mine :-)
>>
>> Verified, and this line solved all the warnings:
>>
>> CC fs/reiserfs/bitmap.o
>> CC fs/reiserfs/do_balan.o
>> CC fs/reiserfs/namei.o
>> CC fs/reiserfs/inode.o
>> CC fs/reiserfs/file.o
>> CC fs/reiserfs/dir.o
>> CC fs/reiserfs/fix_node.o
>> CC fs/reiserfs/super.o
>> CC fs/reiserfs/prints.o
>> CC fs/reiserfs/objectid.o
>> CC fs/reiserfs/lbalance.o
>> CC fs/reiserfs/ibalance.o
>> CC fs/reiserfs/stree.o
>> CC fs/reiserfs/hashes.o
>> CC fs/reiserfs/tail_conversion.o
>> CC fs/reiserfs/journal.o
>> CC fs/reiserfs/resize.o
>> CC fs/reiserfs/item_ops.o
>> CC fs/reiserfs/ioctl.o
>> CC fs/reiserfs/xattr.o
>> CC fs/reiserfs/lock.o
>> CC fs/reiserfs/procfs.o
>> AR fs/reiserfs/built-in.a
>>
>> Just FWIW, back then in year 2000/2001 a journaling file system on my
>> Knoppix box was a quantum leap - it would simply replay the journal if
>> there was a power loss before shutdown. No several minutes of fsck.
>
> Well, there was also ext3 at that time already :-) That's where I became
> familiar with the idea of journalling. Reiserfs was interesting to me
> because of completely different approach to on-disk format (b-tree with
> formatted items), packing of small files / file tails (interesting in 2000,
> not so much 20 years later) and reasonable scalability for large
> directories.
>
>> I think your idea is great and if you wish a patch could be submitted
>> after the merge window.
>
> I'll leave it up to you. If the warnings annoy you, send the patch along
> the lines I've proposed (BTW (void)cond should better be (void)(cond) for
> macro safety) and I'll push it to Linus.
>
> Honza
Hi, Jan,
After a short break, I just tried a full build with this hack against the vanilla
linux-next tree:
#define RFALSE( cond, format, args... ) do { (void)(cond); } while( 0 )
and it breaks at least here:
In file included from fs/reiserfs/do_balan.c:15:
fs/reiserfs/do_balan.c: In function ‘balance_leaf_when_delete_del’:
fs/reiserfs/do_balan.c:86:28: error: ‘ih’ undeclared (first use in this function)
86 | RFALSE(ih_item_len(ih) + IH_SIZE != -tb->insert_size[0],
| ^~
fs/reiserfs/reiserfs.h:919:54: note: in definition of macro ‘RFALSE’
919 | #define RFALSE( cond, format, args... ) do { (void) (cond); } while( 0 )
| ^~~~
./include/linux/byteorder/generic.h:91:21: note: in expansion of macro ‘__le16_to_cpu’
91 | #define le16_to_cpu __le16_to_cpu
| ^~~~~~~~~~~~~
fs/reiserfs/do_balan.c:86:16: note: in expansion of macro ‘ih_item_len’
86 | RFALSE(ih_item_len(ih) + IH_SIZE != -tb->insert_size[0],
| ^~~~~~~~~~~
fs/reiserfs/do_balan.c:86:28: note: each undeclared identifier is reported only once for each function it appears in
86 | RFALSE(ih_item_len(ih) + IH_SIZE != -tb->insert_size[0],
| ^~
fs/reiserfs/reiserfs.h:919:54: note: in definition of macro ‘RFALSE’
919 | #define RFALSE( cond, format, args... ) do { (void) (cond); } while( 0 )
| ^~~~
./include/linux/byteorder/generic.h:91:21: note: in expansion of macro ‘__le16_to_cpu’
91 | #define le16_to_cpu __le16_to_cpu
| ^~~~~~~~~~~~~
fs/reiserfs/do_balan.c:86:16: note: in expansion of macro ‘ih_item_len’
86 | RFALSE(ih_item_len(ih) + IH_SIZE != -tb->insert_size[0],
| ^~~~~~~~~~~
fs/reiserfs/do_balan.c: In function ‘do_balance_starts’:
fs/reiserfs/do_balan.c:1800:16: error: implicit declaration of function ‘check_before_balancing’ [-Werror=implicit-function-declaration]
1800 | RFALSE(check_before_balancing(tb), "PAP-12340: locked buffers in TB");
| ^~~~~~~~~~~~~~~~~~~~~~
fs/reiserfs/reiserfs.h:919:54: note: in definition of macro ‘RFALSE’
919 | #define RFALSE( cond, format, args... ) do { (void) (cond); } while( 0 )
| ^~~~
cc1: some warnings being treated as errors
make[7]: *** [scripts/Makefile.build:244: fs/reiserfs/do_balan.o] Error 1
CC [M] fs/reiserfs/stree.o
In file included from fs/reiserfs/stree.c:15:
fs/reiserfs/stree.c: In function ‘reiserfs_delete_item’:
fs/reiserfs/stree.c:1283:24: error: ‘mode’ undeclared (first use in this function)
1283 | RFALSE(mode != M_DELETE, "PAP-5320: mode must be M_DELETE");
| ^~~~
fs/reiserfs/reiserfs.h:919:54: note: in definition of macro ‘RFALSE’
919 | #define RFALSE( cond, format, args... ) do { (void) (cond); } while( 0 )
| ^~~~
fs/reiserfs/stree.c:1283:24: note: each undeclared identifier is reported only once for each function it appears in
1283 | RFALSE(mode != M_DELETE, "PAP-5320: mode must be M_DELETE");
| ^~~~
fs/reiserfs/reiserfs.h:919:54: note: in definition of macro ‘RFALSE’
919 | #define RFALSE( cond, format, args... ) do { (void) (cond); } while( 0 )
| ^~~~
Last time it compiled, but now it expects variables in (void)(cond) expressions to be defined.
I have try to fix those warnings, submitting the patch for review:
-------------------><---------------------------------------
diff --git a/fs/reiserfs/do_balan.c b/fs/reiserfs/do_balan.c
index 5129efc6f2e6..c8fa3d71ef63 100644
--- a/fs/reiserfs/do_balan.c
+++ b/fs/reiserfs/do_balan.c
@@ -81,11 +81,11 @@ static void balance_leaf_when_delete_del(struct tree_balance *tb)
struct buffer_info bi;
#ifdef CONFIG_REISERFS_CHECK
struct item_head *ih = item_head(tbS0, item_pos);
-#endif
RFALSE(ih_item_len(ih) + IH_SIZE != -tb->insert_size[0],
"vs-12013: mode Delete, insert size %d, ih to be deleted %h",
-tb->insert_size[0], ih);
+#endif
buffer_info_init_tbS0(tb, &bi);
leaf_delete_items(&bi, 0, item_pos, 1, -1);
@@ -1797,8 +1797,8 @@ static inline void do_balance_starts(struct tree_balance *tb)
print_tb(flag, PATH_LAST_POSITION(tb->tb_path),
tb->tb_path->pos_in_item, tb, "check");
*/
- RFALSE(check_before_balancing(tb), "PAP-12340: locked buffers in TB");
#ifdef CONFIG_REISERFS_CHECK
+ RFALSE(check_before_balancing(tb), "PAP-12340: locked buffers in TB");
REISERFS_SB(tb->tb_sb)->cur_tb = tb;
#endif
}
diff --git a/fs/reiserfs/reiserfs.h b/fs/reiserfs/reiserfs.h
index f0e1f29f20ee..027e64853710 100644
--- a/fs/reiserfs/reiserfs.h
+++ b/fs/reiserfs/reiserfs.h
@@ -916,7 +916,7 @@ do { \
#if defined( CONFIG_REISERFS_CHECK )
#define RFALSE(cond, format, args...) __RASSERT(!(cond), "!(" #cond ")", format, ##args)
#else
-#define RFALSE( cond, format, args... ) do {;} while( 0 )
+#define RFALSE( cond, format, args... ) do { (void) (cond); } while( 0 )
#endif
#define CONSTF __attribute_const__
diff --git a/fs/reiserfs/stree.c b/fs/reiserfs/stree.c
index 5faf702f8d15..eed1a461169e 100644
--- a/fs/reiserfs/stree.c
+++ b/fs/reiserfs/stree.c
@@ -1280,7 +1280,9 @@ int reiserfs_delete_item(struct reiserfs_transaction_handle *th,
&del_size,
max_reiserfs_offset(inode));
+#ifdef CONFIG_REISERFS_CHECK
RFALSE(mode != M_DELETE, "PAP-5320: mode must be M_DELETE");
+#endif
copy_item_head(&s_ih, tp_item_head(path));
s_del_balance.insert_size[0] = del_size;
--
Thanks.
Best regards,
Mirsad Todorovac
