[PATCH v3 2/3] erofs-utils: fuse: support symlink & special inode

[Gao Xiang]

Hi Guifu,

On Sat, Dec 05, 2020 at 04:42:04PM +0800, Li GuiFu via Linux-erofs wrote:
> 
> 
> On 2020/11/27 19:46, Gao Xiang via Linux-erofs wrote:
> > From: Huang Jianan <huangjianan at oppo.com>
> > 
> > This patch adds symlink and special inode (e.g. block dev, char,
> > socket, pipe inode) support.
> > 
> > Signed-off-by: Huang Jianan <huangjianan at oppo.com>
> > Signed-off-by: Guo Weichao <guoweichao at oppo.com>
> > Signed-off-by: Gao Xiang <hsiangkao at aol.com>
> > ---
> >  fuse/main.c | 10 ++++++++++
> >  lib/namei.c | 22 ++++++++++++++++++----
> >  2 files changed, 28 insertions(+), 4 deletions(-)
> > 
> 
> It looks good
> Reviewed-by: Li Guifu <bluce.lee at aliyun.com>
> Tested-by: Li Guifu <bluce.lee at aliyun.com>

My tendency is that once a patch is merged to dev branch from
experimental. There is no chance to apply more tags since we need to
stablize the commit ID for dev/master branch (git commit --amend will
modify the original commit ID). So I have to ignore these RVB tags
for the reason above.

And the merging route is "experimental" (can rebase) -> "dev" (cannot
rebase) -> "master".

Also, it's actually hard to carefully review erofsfuse code here with
limited time if your bandwidth is limited as well, so I applied directly
as what I said in
https://lore.kernel.org/r/20201127131630.GA654423@xiangao.remote.csb/

But yeah, more testing is always welcome, and thanks for the testing :)

Finally, my suggestion is that it's better to use erofsfuse rather than
customized ".ko" for older linux x86 versions at HUAWEI as well, since
the old internal ".ko" versions are quite hacky and messy (at least due
to many kernel API changes).

Thanks,
Gao Xiang