Re: [PATCH v3] btrfs: warn about extent buffer that can not be released

[Qu Wenruo <quwenruo.btrfs@gmx.com>]

在 2026/4/28 01:18, David Sterba 写道:
[...]
>>   
>> +static void invalidate_btree_folios(struct btrfs_fs_info *fs_info)
> 
> This is too close to the generic invalidate_inode_pages2, please add
> btrfs_ prefix.

Sure, and I'll try to get a better name for it, as we also have 
btree_invalidate_folio().

[...]
>> --- a/fs/btrfs/transaction.h
>> +++ b/fs/btrfs/transaction.h
>> @@ -255,13 +255,13 @@ do {								\
>>   		__first = true;					\
>>   		if (WARN(btrfs_abort_should_print_stack(error),	\
>>   			KERN_ERR				\
>> -			"BTRFS: Transaction aborted (error %d)\n",	\
>> -			(error))) {					\
>> +			"BTRFS: Transaction %llu aborted (error %d)\n",	\
>> +			(trans)->transid, (error))) {			\
>>   			/* Stack trace printed. */			\
>>   		} else {						\
>>   			btrfs_err((trans)->fs_info,			\
>> -				  "Transaction aborted (error %d)",	\
>> -				  (error));			\
>> +			"Transaction %llu aborted (error %d)",	\
>> +				  (trans)->transid, (error));	\
> 
> Adding the transaction number adds like 4KiB of object code because the
> calls are inlined so we can have exact location and stack.
> 
> It could be possibly moved to __btrfs_abort_transaction() but with some
> additinal shuffling of the code from macro to the handler.
> 

I can remove the transid part, as it doesn't contribute to the final bug 
that got pinned down.

It's mostly a precautious debug feature.

Thanks,
Qu