mbox series

[0/3] squashfs fixes

Message ID 20200715201143.2168632-1-thomas.petazzoni@bootlin.com
Headers show
Series squashfs fixes | expand

Message

Thomas Petazzoni July 15, 2020, 8:11 p.m. UTC
Hello Joao,

As I started reviewing your patch series, I encountered a number of
build issues after applying your patches, and building on the latest
master. The following patches address the build failures I have
encountered. Could you squash them into your commits, so that those
fixes are part of your v2 ? Of course, please do some testing, as I
have only done build testing so far.

Also, even with those fixes in place, I do still get a worrying gcc
warning (from gcc 10):

fs/squashfs/sqfs_inode.c: In function ‘sqfs_find_inode’:
fs/squashfs/sqfs_inode.c:72:34: warning: array subscript l is outside array bounds of ‘struct squashfs_directory_index[0]’ [-Warray-bounds]
   72 |     index_list_size += ldir.index[l].size + 1;
      |                        ~~~~~~~~~~^~~
In file included from fs/squashfs/sqfs_inode.c:16:
fs/squashfs/sqfs_filesystem.h:207:34: note: while referencing ‘index’
  207 |  struct squashfs_directory_index index[0];
      |                                  ^~~~~
fs/squashfs/sqfs_inode.c:31:29: note: defined here ‘ldir’
   31 |  struct squashfs_ldir_inode ldir;
      |                             ^~~~

Could you investigate this ?

Thanks,

Thomas

Thomas Petazzoni (3):
  cmd/sqfs.c, include/fs.h: fix build failures, don't expose do_sqfs_*()
    functions globally
  fs/squashfs: use "struct disk_partition" instead of "disk_partition_t"
  fs/squashfs: do not use CMD_RET_* defines in the filesystem code

 cmd/sqfs.c         |  5 ++--
 fs/squashfs/sqfs.c | 66 ++++++++++++++++++++--------------------------
 include/fs.h       | 12 ---------
 include/squashfs.h |  4 ++-
 4 files changed, 34 insertions(+), 53 deletions(-)

Comments

Joao Marcos Costa July 16, 2020, 7:51 a.m. UTC | #1
Hello, Thomas!

On Wed, 15 Jul 2020 22:11:40 +0200
Thomas Petazzoni <thomas.petazzoni@bootlin.com> wrote:

> Hello Joao,
> 
> As I started reviewing your patch series, I encountered a number of
> build issues after applying your patches, and building on the latest
> master. The following patches address the build failures I have
> encountered. Could you squash them into your commits, so that those
> fixes are part of your v2 ? Of course, please do some testing, as I
> have only done build testing so far.

Thank you for this series, I was actually working on the v2020.04
branch, and I think this explains why I was not seeing such errors. 

> Also, even with those fixes in place, I do still get a worrying gcc
> warning (from gcc 10):

I will update my gcc version, because currently I am using gcc 7.5.

> 
> fs/squashfs/sqfs_inode.c: In function ‘sqfs_find_inode’:
> fs/squashfs/sqfs_inode.c:72:34: warning: array subscript l is outside
> array bounds of ‘struct squashfs_directory_index[0]’ [-Warray-bounds]
> 72 |     index_list_size += ldir.index[l].size + 1;
> |                        ~~~~~~~~~~^~~ In file included from
> fs/squashfs/sqfs_inode.c:16: fs/squashfs/sqfs_filesystem.h:207:34:
> note: while referencing ‘index’ 207 |  struct
> squashfs_directory_index index[0]; |
> ^~~~~ fs/squashfs/sqfs_inode.c:31:29: note: defined here ‘ldir’
>    31 |  struct squashfs_ldir_inode ldir;
>       |                             ^~~~
> 
> Could you investigate this ?

Surely I will, thanks!
 
> Thanks,
> 
> Thomas
> 
> Thomas Petazzoni (3):
>   cmd/sqfs.c, include/fs.h: fix build failures, don't expose
> do_sqfs_*() functions globally
>   fs/squashfs: use "struct disk_partition" instead of
> "disk_partition_t" fs/squashfs: do not use CMD_RET_* defines in the
> filesystem code
> 
>  cmd/sqfs.c         |  5 ++--
>  fs/squashfs/sqfs.c | 66
> ++++++++++++++++++++-------------------------- include/fs.h       |
> 12 --------- include/squashfs.h |  4 ++-
>  4 files changed, 34 insertions(+), 53 deletions(-)
>
Thomas Petazzoni July 16, 2020, 8:33 a.m. UTC | #2
On Thu, 16 Jul 2020 09:51:25 +0200
Joao Marcos Costa <joaomarcos.costa@bootlin.com> wrote:

> > As I started reviewing your patch series, I encountered a number of
> > build issues after applying your patches, and building on the latest
> > master. The following patches address the build failures I have
> > encountered. Could you squash them into your commits, so that those
> > fixes are part of your v2 ? Of course, please do some testing, as I
> > have only done build testing so far.  
> 
> Thank you for this series, I was actually working on the v2020.04
> branch, and I think this explains why I was not seeing such errors. 

OK. When you're submitting patches to U-Boot, since the patches will
very likely be applied on master, the patches you send should be based
on master. Otherwise, the maintainer who will apply your patches will
face the sort of problems that I did face.

> > Also, even with those fixes in place, I do still get a worrying gcc
> > warning (from gcc 10):  
> 
> I will update my gcc version, because currently I am using gcc 7.5.

You could also keep gcc 7.x on your system, but simply use a gcc 10.x
toolchain when cross-compiling U-Boot for ARM for example, as I know
you're also testing this on an ARM platform.

Thomas