Changes to OpenAFS to support FreeBSD 8-CURRENT

[Ben Kaduk]

Hi Boris,

To get a OpenAFS compiling on a couple-day-old current, I needed to
change (in addition to
the vinvalbuf change mentioned elsewhere)

VSUID to S_ISUID and VSGID to S_ISGID in VNOPS/afs_vnop_attrs.c
from this change:
trasz       2008-09-10 13:16:41 UTC
(note that S_ISUID is from stat.h and VUID was from vnode.h)

remove the curthread argument from VOP_GETATTR() (two occurrences)
in FBSD/osi_file.c:143  (this is under AFS_FBSD50_ENV ???)
from this change:
attilio     2008-08-28 15:23:18 UTC

The vinvalbuf change also affects FBSD/osi_vm.c in two lines.

change ap->a_mode to ap->a_accmode in FBSD/osi_vnodeops.c:691
and in the comment above.  Arguably, this might be rather a bug
in the vnode_if.h header where the structure is defined
(libafs/MODLOAD/vnode_if.h, apparently).

Now, everything compiles, but kldload is claiming that libafs.ko
doesn't exist ...

-Ben Kaduk
_______________________________________________
freebsd-afs@... mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-afs
To unsubscribe, send any mail to "freebsd-afs-unsubscribe@..."

[Matt W. Benjamin]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256

Hi Ben,

I got your previous email, but wasn't sure I could get to this.  I
haven't in fact been focused on 8-CURRENT, but on vnode locking and
refcounting correctness across 7.0 to 8-CURRENT.  (Making headway...)

Anyway, I set that aside and whacked through a similar effort.  The
things you got sound right to me, though I'm surprised krx and
afs_server.c compiled, but, it might have to do with whether you have
network stack virtualization enabled.  The  module doesn't load
presumably because of unresolved externals--here either the
aforementioned (like in_ifaddrhead), and probably "suser(thread)" which
is gone, replaced with "priv_check(thread)" (what makes me think that
one won't be around long).

Openafs-1.4.8 (openafs.org) with RT # 123933 compiles and loads, as
above,  but I have not run the code, it's just to block something in, ATM.

Matt

Ben Kaduk wrote:

> Hi Boris,
>
> To get a OpenAFS compiling on a couple-day-old current, I needed to
> change (in addition to
> the vinvalbuf change mentioned elsewhere)
>
> VSUID to S_ISUID and VSGID to S_ISGID in VNOPS/afs_vnop_attrs.c
> from this change:
> trasz       2008-09-10 13:16:41 UTC
> (note that S_ISUID is from stat.h and VUID was from vnode.h)
>
> remove the curthread argument from VOP_GETATTR() (two occurrences)
> in FBSD/osi_file.c:143  (this is under AFS_FBSD50_ENV ???)
> from this change:
> attilio     2008-08-28 15:23:18 UTC
>
> The vinvalbuf change also affects FBSD/osi_vm.c in two lines.
>
> change ap->a_mode to ap->a_accmode in FBSD/osi_vnodeops.c:691
> and in the comment above.  Arguably, this might be rather a bug
> in the vnode_if.h header where the structure is defined
> (libafs/MODLOAD/vnode_if.h, apparently).
>
> Now, everything compiles, but kldload is claiming that libafs.ko
> doesn't exist ...
>
> -Ben Kaduk
> _______________________________________________
> freebsd-afs@... mailing list
> http://lists.freebsd.org/mailman/listinfo/freebsd-afs
> To unsubscribe, send any mail to "freebsd-afs-unsubscribe@..."

- --

Matt Benjamin

The Linux Box
206 South Fifth Ave. Suite 150
Ann Arbor, MI  48104

http://linuxbox.com

tel. [phone]
fax. [phone]
cel. 734-216-5309

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.7 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFJRHXBJiSUUSaRdSURCIEPAJsFcRHTBEDsWs4y2r3h0LJlQ7LjzgCdGJS4
uFYwecbqcgAQeA1bXhSnqG4=
=f20Q
-----END PGP SIGNATURE-----
_______________________________________________
freebsd-afs@... mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-afs
To unsubscribe, send any mail to "freebsd-afs-unsubscribe@..."

[Ben Kaduk]

On Sat, Dec 13, 2008 at 9:56 PM, Matt Benjamin <matt@...> wrote:

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA256
>
> Hi Ben,
>
> I got your previous email, but wasn't sure I could get to this.  I
> haven't in fact been focused on 8-CURRENT, but on vnode locking and
> refcounting correctness across 7.0 to 8-CURRENT.  (Making headway...)
>
> Anyway, I set that aside and whacked through a similar effort.  The
> things you got sound right to me, though I'm surprised krx and
> afs_server.c compiled, but, it might have to do with whether you have
> network stack virtualization enabled.  The  module doesn't load
> presumably because of unresolved externals--here either the
> aforementioned (like in_ifaddrhead), and probably "suser(thread)" which
> is gone, replaced with "priv_check(thread)" (what makes me think that
> one won't be around long).

Right.  After a bit of googling, I thought to check the dmesg, where
the helpful error
link_elf: symbol suser undefined
had appeared.
Now I get to find where kldload is printing "No such file or
directory" instead of something useful, and file a bug.

>
> Openafs-1.4.8 (openafs.org) with RT # 123933 compiles and loads, as
> above,  but I have not run the code, it's just to block something in, ATM.
>

Okay, I'll look at that some more.

Thanks for taking the time out for me.

-Ben Kaduk

> Matt
>
> Ben Kaduk wrote:
>> Hi Boris,
>>
>> To get a OpenAFS compiling on a couple-day-old current, I needed to
>> change (in addition to
>> the vinvalbuf change mentioned elsewhere)
>>
>> VSUID to S_ISUID and VSGID to S_ISGID in VNOPS/afs_vnop_attrs.c
>> from this change:
>> trasz       2008-09-10 13:16:41 UTC
>> (note that S_ISUID is from stat.h and VUID was from vnode.h)
>>
>> remove the curthread argument from VOP_GETATTR() (two occurrences)
>> in FBSD/osi_file.c:143  (this is under AFS_FBSD50_ENV ???)
>> from this change:
>> attilio     2008-08-28 15:23:18 UTC
>>
>> The vinvalbuf change also affects FBSD/osi_vm.c in two lines.
>>
>> change ap->a_mode to ap->a_accmode in FBSD/osi_vnodeops.c:691
>> and in the comment above.  Arguably, this might be rather a bug
>> in the vnode_if.h header where the structure is defined
>> (libafs/MODLOAD/vnode_if.h, apparently).
>>
>> Now, everything compiles, but kldload is claiming that libafs.ko
>> doesn't exist ...
>>
>> -Ben Kaduk
>> _______________________________________________
>> freebsd-afs@... mailing list
>> http://lists.freebsd.org/mailman/listinfo/freebsd-afs
>> To unsubscribe, send any mail to "freebsd-afs-unsubscribe@..."
>
>
>
>
> - --
>
> Matt Benjamin
>
> The Linux Box
> 206 South Fifth Ave. Suite 150
> Ann Arbor, MI  48104
>
> http://linuxbox.com
>
> tel. [phone]
> fax. [phone]
> cel. 734-216-5309
>
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.7 (GNU/Linux)
> Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
>
> iD8DBQFJRHXBJiSUUSaRdSURCIEPAJsFcRHTBEDsWs4y2r3h0LJlQ7LjzgCdGJS4
> uFYwecbqcgAQeA1bXhSnqG4=
> =f20Q
> -----END PGP SIGNATURE-----
>

_______________________________________________
freebsd-afs@... mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-afs
To unsubscribe, send any mail to "freebsd-afs-unsubscribe@..."

[Matt W. Benjamin]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256

RT # 123936 overlays, conditionally includes <sys/priv.h> and passes the
PRIV_ROOT argument as expected

Matt Benjamin wrote:
>
> Openafs-1.4.8 (openafs.org) with RT # 123933 compiles and loads, as
> above,  but I have not run the code, it's just to block something in, ATM.

- --

Matt Benjamin

The Linux Box
206 South Fifth Ave. Suite 150
Ann Arbor, MI  48104

http://linuxbox.com

tel. [phone]
fax. [phone]
cel. 734-216-5309

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.7 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFJRahvJiSUUSaRdSURCF5+AJ9Ng0tzOYtFWsKiR08aDJkpCPiSTQCeIaaD
ShQf6mK85MP/n4O/dhTuIOs=
=bfj3
-----END PGP SIGNATURE-----
_______________________________________________
freebsd-afs@... mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-afs
To unsubscribe, send any mail to "freebsd-afs-unsubscribe@..."

[Ben Kaduk]

On Sat, Dec 13, 2008 at 9:56 PM, Matt Benjamin <matt@...> wrote:

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA256
>
> Hi Ben,
>
> I got your previous email, but wasn't sure I could get to this.  I
> haven't in fact been focused on 8-CURRENT, but on vnode locking and
> refcounting correctness across 7.0 to 8-CURRENT.  (Making headway...)
>
> Anyway, I set that aside and whacked through a similar effort.  The
> things you got sound right to me, though I'm surprised krx and
> afs_server.c compiled, but, it might have to do with whether you have
> network stack virtualization enabled.  The  module doesn't load
> presumably because of unresolved externals--here either the
> aforementioned (like in_ifaddrhead), and probably "suser(thread)" which
> is gone, replaced with "priv_check(thread)" (what makes me think that
> one won't be around long).
>
> Openafs-1.4.8 (openafs.org) with RT # 123933 compiles and loads, as
> above,  but I have not run the code, it's just to block something in, ATM.
>

It's possible that I'm doing something wrong, but I'm getting panics with
returning from a syscall with a lock held.  I have a project that I need to
finish up for my classes, here, so I shouldn't spend more time on this for
a few days.  If anyone really wants a core before then, let me know, but
I'm pretty sure that my libafs.ko doesn't have debugging symbols,
so it's probably not worth it.

Just FYI,

Ben Kaduk
_______________________________________________
freebsd-afs@... mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-afs
To unsubscribe, send any mail to "freebsd-afs-unsubscribe@..."