Re: CVS commit: src/sbin/shutdown

[Robert Elz]

Date:Tue, 13 Sep 2016 15:25:26 +0700
From:Robert Elz 
Message-ID:  <27729.1473755...@andromeda.noi.kre.to>

  | Shutdown originated in 6th edn (maybe 5th) UNSW (pdp-11) unix,
  | 
  | At one stage it had (in comments in the source) our names all neatly
  | lined up ...
  | 
  | Ian Johnstone 
  | K. Robert Elz
  | William N Joy

Turns out I didn't remember the format as well as I thought, it was
actually ...

 *  Ian Johnstone, Sydney, 1977
 *  Robert Elz, Melbourne, 1978
 *  Peter Lamb, Melbourne, 1980
 *  William Joy, Berkeley, 1981 
 *  Michael Toy, Berkeley, 1981
 *  Dave Presotto, Berkeley, 1981

The neat alignment I remembered was still there though, and the final
mis-alignment (and then this stuff was all deleted a little later).

It was added Feb 1981, 4.1BSD was June 1981, so shutdown was in 4.1
(and the man page SCCS file confirms that as well, the 4.1 and 4.2 man
pages both appear there.)

  | till someone else in CSRG (one of the Keith's I suspect) modified it more,

Keith Bostic did get at it, the log says "complete rewrite", but that
was later, 1988, after 4.3, so would have been for Tahoe or Reno.

kre

Re: CVS commit: src/usr.bin/config

[Roy Marples]

On 09/09/2016 22:09, Christos Zoulas wrote:
> Module Name:  src
> Committed By: christos
> Date: Fri Sep  9 21:09:11 UTC 2016
> 
> Modified Files:
>   src/usr.bin/config: defs.h files.c sem.c
> 
> Log Message:
> Make attribute deselection work:
> - when deselecting attributes, remove files that depend on them
> - when deselecting attributes, remove devices that depend on them
> 
> 
> To generate a diff of this commit:
> cvs rdiff -u -r1.97 -r1.98 src/usr.bin/config/defs.h
> cvs rdiff -u -r1.35 -r1.36 src/usr.bin/config/files.c
> cvs rdiff -u -r1.75 -r1.76 src/usr.bin/config/sem.c

#   compile  config/sem.o
/home/roy/tools/amd64/bin/x86_64--netbsd-clang -O2 -g -fPIE
-fstack-protector -Wstack-protector  --param ssp-buffer-size=1
-std=gnu99 -Wno-sign-compare -Wno-pointer-sign  -Wall
-Wstrict-prototypes -Wmissing-prototypes -Wpointer-arith
-Wno-sign-compare  -Wa,--fatal-warnings -Wreturn-type -Wswitch -Wshadow
-Wcast-
qual -Wwrite-strings -Wextra -Wno-unused-parameter -Wno-sign-compare
-Wold-style-definition -Wconversion -Wsign-compare -Wformat=2
-Wpointer-sign -Wmissing-no
return  -Werror -Wno-format-y2k  -pipe
--sysroot=/home/roy/src/src/obj/destdir.amd64
-I/home/roy/src/src/usr.bin/config -I. -I/home/roy/src/src/usr.bin/cksum
-D_FORTIFY_SOURCE=2 -c/home/roy/src/src/usr.bin/config/sem.c
/home/roy/src/src/usr.bin/config/sem.c:351:29: error: variable 'a' is
uninitialized when used here [-Werror,-Wuninitialized]
addvalue(V_ATTRIBUTE, a, dep));
  ^
/home/roy/src/src/usr.bin/config/defs.h:528:68: note: expanded from
macro 'ht_insert2'
#define ht_insert2(ht, nam1, nam2, val) ht_insrep2(ht, nam1, nam2, val, 0)
   ^~~
/home/roy/src/src/usr.bin/config/sem.c:336:16: note: initialize the
variable 'a' to silence this warning
struct attr *a, *dep;
  ^
   = NULL
1 error generated.
*** [sem.o] Error code 1

Re: CVS commit: src/sys

[Roy Marples]

On 13/09/2016 02:28, Ryota Ozaki wrote:
> On Mon, Sep 12, 2016 at 6:28 PM, Kengo NAKAHARA  wrote:
>> Hi,
>>
>> On 2016/09/12 17:28, Roy Marples wrote:
>>> On 09/09/2016 20:08, Martin Husemann wrote:
 On Thu, Sep 08, 2016 at 02:19:15PM +0900, Kengo NAKAHARA wrote:
>
> On 2016/09/08 0:41, Roy Marples wrote:
>> Module Name:   src
>> Committed By:  roy
>> Date:  Wed Sep  7 15:41:44 UTC 2016
>>
>> Modified Files:
>>src/sys/netinet: ip_input.c
>>src/sys/netinet6: ip6_input.c
>>
>> Log Message:
>> Disallow input to detached addresses because they are not yet valid.
>>
>>
>> To generate a diff of this commit:
>> cvs rdiff -u -r1.340 -r1.341 src/sys/netinet/ip_input.c
>> cvs rdiff -u -r1.167 -r1.168 src/sys/netinet6/ip6_input.c
>>
>> Please note that diffs are not public domain; they are subject to the
>> copyright notices on the relevant files.
>
> ATF net/if_pppoe/t_pppoe:pap fails after this commit. I revert this
> commit locally, and then the ATF become successful.

 I see the failure too, but manually configuring two rump servers in a very
 similar setup than the test still works for me - something must be wrong
 in the test, but I don't see it right now.
>>>
>>> martin@ did more testing and discovered that the PPP connection works
>>> fine, so something is wrong in the test.
>>>
>>> I believe that the function wait_for_session_established needs an
>>> additional pause to wait for the detached flag to clear.
>>> If the interface is marked IFF_UP at this point, it should be possible
>>> to add ifconfig -w 5 -W -5 and the test should then succeed. I'm getting
>>> symbol errors still trying to run this test so cannot fix it myself
>>> right now.
>>
>> I add ifconfig -w 15 -W 15 in wait_for_session_established(), however
>> net/if_pppoe/t_pppoe:pap still fails.
>>
>> Hmm, does my below modification match your pointing out, doesn't it?
> 
> It seems that DAD doesn't run on ifconfig pppoe0 up
> and pppoe0 remains as detached.
> 
> The following processing is happening:
> - ifconfig up calls ioctl(SIOCSIFFLAGS)
>   that calls ifioctl_common normally
> - ifioctl_common schedules DAD eventually
>   but only if IFF_UP & IFF_RUNNING
>   (see in_if_link_up)
> - However, some interfaces including pppoe set
>   IFF_RUNNING *after* calling ifioctl_common
> - Such interfaces don't run DAD
>   and remain as detached
> 
> I don't know how to fix it properly.

OK, I've had a quick look over the code.
It looks like IFF_RUNNING is used to indicate whether a PPP session is
currently running or not as opposed to hardware resources being allocated.

It strikes me that the easiest fix is to set IFF_RUNNING before calling
ifioctl_common if going_up and clear it if there is an error.

We should also hook the interface link state code into ppp as well
because we should know when the link is ready for use.

Any reason why this cannot be done?

Roy

Re: CVS commit: src/sys

[Martin Husemann]

On Tue, Sep 13, 2016 at 01:38:38PM +0100, Roy Marples wrote:
> Where does pppoe even set IFF_RUNNING? I couldn't find that.

Various places in if_spppsubr.c, depending on the interface flags for
incoming or "auto dial".

Martin

Re: CVS commit: src/sys

[Roy Marples]

On 13/09/2016 02:28, Ryota Ozaki wrote:
> It seems that DAD doesn't run on ifconfig pppoe0 up
> and pppoe0 remains as detached.
> 
> The following processing is happening:
> - ifconfig up calls ioctl(SIOCSIFFLAGS)
>   that calls ifioctl_common normally
> - ifioctl_common schedules DAD eventually
>   but only if IFF_UP & IFF_RUNNING
>   (see in_if_link_up)
> - However, some interfaces including pppoe set
>   IFF_RUNNING *after* calling ifioctl_common
> - Such interfaces don't run DAD
>   and remain as detached
> 
> I don't know how to fix it properly.

Where does pppoe even set IFF_RUNNING? I couldn't find that.

Re: CVS commit: src/sbin/shutdown

[Robert Elz]

Date:Mon, 12 Sep 2016 06:22:35 +
From:"Sevan Janiyan" 
Message-ID:  <20160912062235.6396bf...@cvs.netbsd.org>

  | Modified Files:
  | src/sbin/shutdown: shutdown.8
  | 
  | Log Message:
  | shutdown appeared as a standalone utility in 4.2BSD release, in 4BSD its a
  | target in init.

I will need to find where I put my CSRG SCCS files to check, but I don't
think shutdown can really have first appeared in 4.2

Shutdown originated in 6th edn (maybe 5th) UNSW (pdp-11) unix, originally
written by Ian Johnstone - I had it (and modified it) in the Melbourne Uni
version of Wollongong University's Interdata 6th edition, and then added
to 3BSD at Melb, and then added it to the Berkeley sources, where Bill Joy
modified it even more.

At one stage it had (in comments in the source) our names all neatly
lined up ...

Ian Johnstone 
K. Robert Elz
William N Joy

till someone else in CSRG (one of the Keith's I suspect) modified it more,
and added their name, and the nice block layout was broken ... so all that
nonsense was removed.

But because Bill worked on this, it really cannot have appeared in BSD
later than 4.1 (and the original was probably correct - 4BSD.)

The actual shutdown was always a "target in init", through all of this,
shutdown's job was to notify users of the upcoming outage, install an
/etc/nologin at the appropriate time, and then poke init - that is, it
provided orderly planned shutdown/reboot - init on the other hand just
"shutdown now, immediately".

kre

Re: CVS commit: src/sbin/newfs

[Robert Elz]

Date:Tue, 13 Sep 2016 00:50:30 +0100
From:Sevan Janiyan 
Message-ID:  <2d30ad57-2341-3b08-1d78-32b468cd5...@netbsd.org>

  | I can't seem to find the 4.3BSD-Reno release notes at the moment, but
  | the convention is interim releases do not count.
  | There are examples of that where 4.2BSD is being referenced in inherited
  | man pages when in-fact the tools actually showed up in 4.1{a,b,c}.

I think it is reasonable to ignore 4.1a (etc) for this kind of purpose,
but as I recall, 4.3 Reno & Tahoe were really full releases

kre