[PATCH] erofs-utils: do not check ->idata_size for compressed files in erofs_prepare_inode_buffer()

[Gao Xiang]

On Wed, Sep 01, 2021 at 10:20:38AM +0800, Yue Hu wrote:
> On Tue, 31 Aug 2021 20:20:11 +0800
> Gao Xiang <hsiangkao at linux.alibaba.com> wrote:
> 
> > On Tue, Aug 31, 2021 at 07:56:14PM +0800, Yue Hu wrote:
> > > On Tue, 31 Aug 2021 19:15:50 +0800
> > > Gao Xiang <hsiangkao at linux.alibaba.com> wrote:  
> > 
> > ...
> > 
> > > > > > > > >         
> > > > > > > > > > 
> > > > > > > > > > BTW, if you have some interest, would you like to implement it? :)        
> > > > > > > > > 
> > > > > > > > > I don't know if i can finish it. But i would like to have a try :)        
> > > > > > > > 
> > > > > > > > My rough thought is to try to inline the last tail compresseed
> > > > > > > > extent after the on-disk compressed extents, maybe we could let
> > > > > > > > it work for non-compact (legacy) compress index format cases...        
> > > > > > > 
> > > > > > > I mean try to implement non-compact (legacy) compress index format cases
> > > > > > > first.    
> > > > > 
> > > > > I'm trying to do it under 4.19 code (since i have no 5.x environment temporarily).
> > > > > 
> > > > > Now, i think mkfs should be done. But, kernel side seems not working fine(no crash,
> > > > > no decompression warning/bug). Only some files are working, others not. I'm sure i
> > > > > can catch the inline data correctly via file dump. And I'm trying debug the issue.
> > > > > Maybe i need more time to read/understand more decompression code related.
> > > > > 
> > > > > BTW, now i understand no need to go z_erofs_vle_work_xxx for inline part(cur-end)
> > > > > , just go next_part after mapping as below, am i right? 
> > > > >     
> > > > 
> > > > You are right. For the common cases (except for fiemap or cases to get the exact
> > > > decompressed length), we only need to calculate the start of the compression extent,
> > > > so it's transversal in the reverse order.
> > > > 
> > > > But really... Again, I don't suggest using 4.19 staging code for real production
> > > > or further development. The uncompressed part is considered as stable, but
> > > > compression side may not (also it was disabled by default). Please also see,
> > > > https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/staging/erofs/Kconfig?h=v4.19#n86
> > > > 
> > > > " config EROFS_FS_ZIP
> > > >   bool "EROFS Data Compresssion Support"
> > > >   depends on EROFS_FS
> > > >   help
> > > >     Currently we support VLE Compression only.
> > > >     Play at your own risk.
> > > > 
> > > >     If you don't want to use compression feature, say N. "
> > > > 
> > > > Our original first real production codebase was between 5.2~5.3. Therefore,
> > > > I suggest using >= 5.4 LTS codebase for production. You could also find
> > > > some backport codebase on github, e.g.:
> > > > https://github.com/nolange/erofs_kernel_4_19
> > > > , which backports 5.6 erofs codebase to 4.19.
> > > > 
> > > > As for tail-packing inline extent feature, how about focusing on on-disk
> > > > design and mkfs/erofsfuse implementation first as PoC?
> > > > 
> > > > I'm afraid that if you only focus on 4.19 codebase, the format of compact
> > > > indexes will be ignored, but "compact indexes" is the default option for
> > > > erofs now since it has less metadata overhead than non-compact indexes,
> > > > so both the sequential / random read are better.  
> > > 
> > > OK, let me develop it under 5.4. I need taking time to find it:)  
> > 
> > As the first step of kernel development, I think using x86 qemu should
> > be better since it's easier to debug than on the embedded device.
> 
> Agree.
> 
> > 
> > For this feature, I'm very glad to discuss some on-disk format first.
> > Since it's not trivial for compact indexes since it's impossible to mark
> > tailing-packing extent with some special blkaddr like non-compact
> > indexes.
> 
> Yes, blkaddr should be an issue for inline case. I can feel that faintly.

Yeah, my point is that "don't mark tail-packing extent since it's somewhat
unfriendly to compact indexes."

> 
> > 
> > My rough thought about this is "to add some new feature flag to "struct
> > z_erofs_map_header" and trigger z_erofs_map_blocks(i_size - 1); at a
> > proper time to get all information about the last tail-packing
> > compression extent", and when submitting io, we erofs_get_meta_page()
> > instead and fill the compressed pages.
> 
> Firstly, I need to add code about inline part to verify my understanding. I
> think i did it almost about what i want to know including z_erofs_map_blocks()
> since i can catch the inline data which is key point for me although kernel side
> does not work fine totally.

Ok, for z_erofs_map_blocks part, erofs-utils and erofs are similiar, I
think maybe debugging erofs-utils makes the life easier...

> 
> Then i can re-factor/re-write it based on that. Yes, i will switch it on >=5.4
> to continue developing later.
> 
> I also think we need a new flag for inline case. I'm just not focus on the flag
> due to my working step above.

Yeah, my suggestion is adding a new flag in "struct z_erofs_map_header".
And reading the last extent when initializing inode or first read.

> 
> Now, i think i can check it about where to add the new flag more proper. Let me
> check it also for your thought mentioned above.
> 
> > 
> > But anyway, I still think focusing on mkfs.erofs and erofsfuse is a good
> > start for this.
> 
> Yes, we should check the compression firstly.
> 
> One more question:
> 
> There's a piece of code (as below) to handle small output size(< PAGE_SIZE) which looks
> like for inline part in z_erofs_decompress_generic()? If so, we also need to go vle 
> decompression flow for inline data just like other data case?
> 
> ```code
> 	if (rq->outputsize <= PAGE_SIZE * 7 / 8) {
> 		dst = erofs_get_pcpubuf(0);
> 		if (IS_ERR(dst))
> 			return PTR_ERR(dst);
> 
> 		rq->inplace_io = false;
> 		ret = alg->decompress(rq, dst);
> 		if (!ret)
> 			copy_from_pcpubuf(rq->out, dst, rq->pageofs_out,
> 					  rq->outputsize);
> 
> 		erofs_put_pcpubuf(dst);
> 		return ret;
> 	}
> ```

Hmmmm.... Nope, this is mainly used to avoid vmap or copy if total output
size is smaller than PAGE_SIZE for inplace I/O.

Considering the following cases, if we read compressed data into a file
page and inplace decompression is _not_ feasible due to lack of safe
inplace margin so we must copy out compressed data in advance.

Generally, for the case above we copy compressed data to some per-CPU
buffer and then do vmap for all needed file pages. But if decompressed
length is less than one page, we could decompress into per-CPU buffer and
copy decompressed data into file page instead, which is faster than the
generic way and is measurable.

Thanks,
Gao Xiang

> 
> Thanks.
> 
> > 
> > Thanks,
> > Gao Xiang
> > 
> > > 
> > > Thanks.
> > >   
> > > > 
> > > > Thanks,
> > > > Gao Xiang
> > > >   
> > > > > Thanks.
> > > > >       
> > > > > > 
> > > > > > Ok, let me try to implement it.
> > > > > > 
> > > > > > Thanks.    
>