Re: [PATCH v2 2/2] fsverity: add tracepoints

Re: [PATCH v2 2/2] fsverity: add tracepoints

[Eric Biggers]

[+Cc fsverity@lists.linux.dev]

On Tue, Jan 20, 2026 at 04:39:17PM -0800, Eric Biggers wrote:
> On Mon, Jan 19, 2026 at 05:56:43PM +0100, Andrey Albershteyn wrote:
> [...]
> > +	TP_printk("ino %lu data size %llu tree size %llu block size %u levels %u",
> [...]
> > +	TP_printk("ino %lu levels %d block_size %d tree_size %lld root_hash %s digest %s",
> 
> Would be nice to make these consistent.  3 of the parameters are the
> same, but the naming and order differs slightly.
> 
> [...]
> > +	TP_printk("ino %lu pos %lld merkle_blocksize %u",
> > +		(unsigned long) __entry->ino,
> > +		__entry->data_pos,
> > +		__entry->block_size)
> 
> Likewise here.  So now we have "block size", "block_size", and
> "merkle_blocksize", all for the same thing.
> 
> > +	TP_printk("ino %lu data_pos %llu hblock_idx %lu level %u hidx %u",
> > +		(unsigned long) __entry->ino,
> 
> And here's data_pos as a %llu, whereas in the previous tracepoint it's
> just pos as an %lld.
> 
> > +TRACE_EVENT(fsverity_verify_merkle_block,
> > +	TP_PROTO(const struct inode *inode, unsigned long index,
> > +		 unsigned int level, unsigned int hidx),
> 
> And the 'index' here is what the previous one calls 'hblock_idx'.
> 
> I think consistent naming would be helpful for people trying to use
> these tracepoints.

Andrey, let me know if you're planning to send a new version with the
naming cleaned up, or if I should do it in a follow-up patch instead.

Christian, can you let me know if it's okay if I take this series
through the fsverity tree, or do you want it to go through the VFS tree?

Thanks,

- Eric

Re: [PATCH v2 2/2] fsverity: add tracepoints

[Andrey Albershteyn]

On 2026-01-24 10:49:54, Eric Biggers wrote:
> [+Cc fsverity@lists.linux.dev]

ops, wrong autocomplete, thanks!

> 
> On Tue, Jan 20, 2026 at 04:39:17PM -0800, Eric Biggers wrote:
> > On Mon, Jan 19, 2026 at 05:56:43PM +0100, Andrey Albershteyn wrote:
> > [...]
> > > +	TP_printk("ino %lu data size %llu tree size %llu block size %u levels %u",
> > [...]
> > > +	TP_printk("ino %lu levels %d block_size %d tree_size %lld root_hash %s digest %s",
> > 
> > Would be nice to make these consistent.  3 of the parameters are the
> > same, but the naming and order differs slightly.
> > 
> > [...]
> > > +	TP_printk("ino %lu pos %lld merkle_blocksize %u",
> > > +		(unsigned long) __entry->ino,
> > > +		__entry->data_pos,
> > > +		__entry->block_size)
> > 
> > Likewise here.  So now we have "block size", "block_size", and
> > "merkle_blocksize", all for the same thing.
> > 
> > > +	TP_printk("ino %lu data_pos %llu hblock_idx %lu level %u hidx %u",
> > > +		(unsigned long) __entry->ino,
> > 
> > And here's data_pos as a %llu, whereas in the previous tracepoint it's
> > just pos as an %lld.
> > 
> > > +TRACE_EVENT(fsverity_verify_merkle_block,
> > > +	TP_PROTO(const struct inode *inode, unsigned long index,
> > > +		 unsigned int level, unsigned int hidx),
> > 
> > And the 'index' here is what the previous one calls 'hblock_idx'.
> > 
> > I think consistent naming would be helpful for people trying to use
> > these tracepoints.
> 
> Andrey, let me know if you're planning to send a new version with the
> naming cleaned up, or if I should do it in a follow-up patch instead.

I will send v2 with renames

-- 
- Andrey

Re: [PATCH v2 2/2] fsverity: add tracepoints

[Christian Brauner]

On Sat, Jan 24, 2026 at 10:49:54AM -0800, Eric Biggers wrote:
> [+Cc fsverity@lists.linux.dev]
> 
> On Tue, Jan 20, 2026 at 04:39:17PM -0800, Eric Biggers wrote:
> > On Mon, Jan 19, 2026 at 05:56:43PM +0100, Andrey Albershteyn wrote:
> > [...]
> > > +	TP_printk("ino %lu data size %llu tree size %llu block size %u levels %u",
> > [...]
> > > +	TP_printk("ino %lu levels %d block_size %d tree_size %lld root_hash %s digest %s",
> > 
> > Would be nice to make these consistent.  3 of the parameters are the
> > same, but the naming and order differs slightly.
> > 
> > [...]
> > > +	TP_printk("ino %lu pos %lld merkle_blocksize %u",
> > > +		(unsigned long) __entry->ino,
> > > +		__entry->data_pos,
> > > +		__entry->block_size)
> > 
> > Likewise here.  So now we have "block size", "block_size", and
> > "merkle_blocksize", all for the same thing.
> > 
> > > +	TP_printk("ino %lu data_pos %llu hblock_idx %lu level %u hidx %u",
> > > +		(unsigned long) __entry->ino,
> > 
> > And here's data_pos as a %llu, whereas in the previous tracepoint it's
> > just pos as an %lld.
> > 
> > > +TRACE_EVENT(fsverity_verify_merkle_block,
> > > +	TP_PROTO(const struct inode *inode, unsigned long index,
> > > +		 unsigned int level, unsigned int hidx),
> > 
> > And the 'index' here is what the previous one calls 'hblock_idx'.
> > 
> > I think consistent naming would be helpful for people trying to use
> > these tracepoints.
> 
> Andrey, let me know if you're planning to send a new version with the
> naming cleaned up, or if I should do it in a follow-up patch instead.
> 
> Christian, can you let me know if it's okay if I take this series
> through the fsverity tree, or do you want it to go through the VFS tree?

Yeah, I can take them no biggie.
Tracepoints, heh, who would've thought.