Re: linux-next: Tree for Jul 18 (header build error)

Re: linux-next: Tree for Jul 18 (header build error)

[Randy Dunlap]

On 7/17/19 8:37 PM, Stephen Rothwell wrote:
> Hi all,
> 
> Please do not add v5.4 material to your linux-next included branches
> until after v5.3-rc1 has been released.
> 
> Changes since 20190717:
> 

on x86_64, when CONFIG_BLOCK is not set:

  CC      include/linux/iomap.h.s
In file included from <command-line>:0:0:
./../include/linux/iomap.h: In function ‘iomap_sector’:
./../include/linux/iomap.h:76:48: error: ‘SECTOR_SHIFT’ undeclared (first use in this function); did you mean ‘SECTIONS_SHIFT’?
  return (iomap->addr + pos - iomap->offset) >> SECTOR_SHIFT;
                                                ^~~~~~~~~~~~


-- 
~Randy

Re: linux-next: Tree for Jul 18 (header build error)

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 772 bytes --]

Hi all,

On Thu, 18 Jul 2019 10:00:22 -0700 Randy Dunlap <rdunlap@infradead.org> wrote:
>
> on x86_64, when CONFIG_BLOCK is not set:
> 
>   CC      include/linux/iomap.h.s
> In file included from <command-line>:0:0:
> ./../include/linux/iomap.h: In function ‘iomap_sector’:
> ./../include/linux/iomap.h:76:48: error: ‘SECTOR_SHIFT’ undeclared (first use in this function); did you mean ‘SECTIONS_SHIFT’?
>   return (iomap->addr + pos - iomap->offset) >> SECTOR_SHIFT;
>                                                 ^~~~~~~~~~~~

include/linux/iomap.h should only be used when CONFIG_BLOCK is set (if
you follow the Kconfig trail).  So maybe this header should only be
compile tested if CONFIG_BLOCK is set.

-- 
Cheers,
Stephen Rothwell

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

Re: linux-next: Tree for Jul 18 (header build error)

[Darrick J. Wong]

On Fri, Jul 19, 2019 at 10:05:57AM +1000, Stephen Rothwell wrote:
> Hi all,
> 
> On Thu, 18 Jul 2019 10:00:22 -0700 Randy Dunlap <rdunlap@infradead.org> wrote:
> >
> > on x86_64, when CONFIG_BLOCK is not set:
> > 
> >   CC      include/linux/iomap.h.s
> > In file included from <command-line>:0:0:
> > ./../include/linux/iomap.h: In function ‘iomap_sector’:
> > ./../include/linux/iomap.h:76:48: error: ‘SECTOR_SHIFT’ undeclared (first use in this function); did you mean ‘SECTIONS_SHIFT’?
> >   return (iomap->addr + pos - iomap->offset) >> SECTOR_SHIFT;
> >                                                 ^~~~~~~~~~~~
> 
> include/linux/iomap.h should only be used when CONFIG_BLOCK is set (if
> you follow the Kconfig trail).  So maybe this header should only be
> compile tested if CONFIG_BLOCK is set.

Yeah, that's basically what Christoph said earlier today.

Granted, nobody replied to my question about where in the kernel would
someone be using iomap to map file offsets to disk locations without
block devices, but ...

... oh, this is some weird "mash all the kernel headers together and see
if they compile" thing, isn't it?  Um, yes, iomap.h should only be
tested if CONFIG_IOMAP=y (which in turn requires CONFIG_BLOCK=y).

--D

> -- 
> Cheers,
> Stephen Rothwell

Re: linux-next: Tree for Jul 18 (header build error)

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 455 bytes --]

Hi Darrick,

On Thu, 18 Jul 2019 17:30:45 -0700 "Darrick J. Wong" <darrick.wong@oracle.com> wrote:
>
> ... oh, this is some weird "mash all the kernel headers together and see
> if they compile" thing, isn't it?  Um, yes, iomap.h should only be
> tested if CONFIG_IOMAP=y (which in turn requires CONFIG_BLOCK=y)

Actually it tries to build each header file on its own to see if it
include all its dependencies.

-- 
Cheers,
Stephen Rothwell

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

Re: linux-next: Tree for Jul 18 (header build error)

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 979 bytes --]

Hi all,

On Fri, 19 Jul 2019 10:05:57 +1000 Stephen Rothwell <sfr@canb.auug.org.au> wrote:
>
> On Thu, 18 Jul 2019 10:00:22 -0700 Randy Dunlap <rdunlap@infradead.org> wrote:
> >
> > on x86_64, when CONFIG_BLOCK is not set:
> > 
> >   CC      include/linux/iomap.h.s
> > In file included from <command-line>:0:0:
> > ./../include/linux/iomap.h: In function ‘iomap_sector’:
> > ./../include/linux/iomap.h:76:48: error: ‘SECTOR_SHIFT’ undeclared (first use in this function); did you mean ‘SECTIONS_SHIFT’?
> >   return (iomap->addr + pos - iomap->offset) >> SECTOR_SHIFT;
> >                                                 ^~~~~~~~~~~~  
> 
> include/linux/iomap.h should only be used when CONFIG_BLOCK is set (if
> you follow the Kconfig trail).  So maybe this header should only be
> compile tested if CONFIG_BLOCK is set.

Or maybe fs/iomap/ needs to grow a private header file for stuff like
iomap_sector() ?

-- 
Cheers,
Stephen Rothwell

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 488 bytes --]