Re: [PATCH 6/8] erofs: get rid of fe->backmost for cache decompression

[Yue Hu]

On Fri, 18 Aug 2023 15:48:08 +0800
Gao Xiang <hsiangkao@xxxxxxxxxxxxxxxxx> wrote:

> On 2023/8/18 13:51, Yue Hu wrote:
> > On Thu, 17 Aug 2023 16:28:11 +0800
> > Gao Xiang <hsiangkao@xxxxxxxxxxxxxxxxx> wrote:
> >   
> >> EROFS_MAP_FULL_MAPPED is more accurate to decide if caching the last
> >> incomplete pcluster for later read or not.
> >>
> >> Signed-off-by: Gao Xiang <hsiangkao@xxxxxxxxxxxxxxxxx>
> >> ---
> >>   fs/erofs/zdata.c | 7 ++-----
> >>   1 file changed, 2 insertions(+), 5 deletions(-)
> >>
> >> diff --git a/fs/erofs/zdata.c b/fs/erofs/zdata.c
> >> index 4009283944ca..c28945532a02 100644
> >> --- a/fs/erofs/zdata.c
> >> +++ b/fs/erofs/zdata.c
> >> @@ -528,8 +528,6 @@ struct z_erofs_decompress_frontend {
> >>   	z_erofs_next_pcluster_t owned_head;
> >>   	enum z_erofs_pclustermode mode;
> >>   
> >> -	/* used for applying cache strategy on the fly */
> >> -	bool backmost;
> >>   	erofs_off_t headoffset;
> >>   
> >>   	/* a pointer used to pick up inplace I/O pages */
> >> @@ -538,7 +536,7 @@ struct z_erofs_decompress_frontend {
> >>   
> >>   #define DECOMPRESS_FRONTEND_INIT(__i) { \
> >>   	.inode = __i, .owned_head = Z_EROFS_PCLUSTER_TAIL, \
> >> -	.mode = Z_EROFS_PCLUSTER_FOLLOWED, .backmost = true }
> >> +	.mode = Z_EROFS_PCLUSTER_FOLLOWED }
> >>   
> >>   static bool z_erofs_should_alloc_cache(struct z_erofs_decompress_frontend *fe)
> >>   {
> >> @@ -547,7 +545,7 @@ static bool z_erofs_should_alloc_cache(struct z_erofs_decompress_frontend *fe)
> >>   	if (cachestrategy <= EROFS_ZIP_CACHE_DISABLED)
> >>   		return false;
> >>   
> >> -	if (fe->backmost)
> >> +	if (!(fe->map.m_flags & EROFS_MAP_FULL_MAPPED))  
> > 
> > So, i understand (map.m_flags & EROFS_MAP_FULL_MAPPED) should be false if allocate cache is needed
> > (fe->backmost is true)?  
> 
> fe->backmost is inaccurate compared with !EROFS_MAP_FULL_MAPPED,
> if !EROFS_MAP_FULL_MAPPED, it should be cached instead.

Okay.

Reviewed-by: Yue Hu <huyue2@xxxxxxxxxxx>

> 
> Thanks,
> Gao Xiang