nftables-goes-automake patch series

nftables-goes-automake patch series

[Jan Engelhardt]

Not much more to elaborate on, this was straight forward to do
and only a matter of time, and really needed :)

 20 files changed, 145 insertions(+), 455 deletions(-)

It is available too from  git://dev.medozas.de/nftables master

Awaiting comments.

Re: nftables-goes-automake patch series

[Patrick McHardy]

Jan Engelhardt wrote:
> Not much more to elaborate on, this was straight forward to do
> and only a matter of time, and really needed :)
>
>  20 files changed, 145 insertions(+), 455 deletions(-)
>
> It is available too from  git://dev.medozas.de/nftables master
>
> Awaiting comments.

Patch 1 looks fine, although I might take the liberty to rename the
file again :)

About automake - I'm glad you're fixing up the makefiles, but I'm
quite attached to the silent build and the automake version that
comes with Debian unstable is 1.10, so I assume it won't be
silent anymore. It also doesn't build anymore on my system, the
parser.h file is not generated:

make  all-recursive
make[1]: Entering directory `/tmp/nftables'
Making all in doc
make[2]: Entering directory `/tmp/nftables/doc'
make[2]: Für das Ziel »all« ist nichts zu tun.
make[2]: Leaving directory `/tmp/nftables/doc'
Making all in files/nftables
make[2]: Entering directory `/tmp/nftables/files/nftables'
make[2]: Für das Ziel »all« ist nichts zu tun.
make[2]: Leaving directory `/tmp/nftables/files/nftables'
Making all in src
make[2]: Entering directory `/tmp/nftables/src'
depbase=`echo scanner.o | sed 's|[^/]*$|.deps/&|;s|\.o$||'`;\
        gcc -DHAVE_CONFIG_H -I. -I..    -D_FILE_OFFSET_BITS=64 
-D_REENTRANT -Wall -Waggregate-return    -Wbad-function-cast 
-Wcast-align -Wdeclaration-after-statement  -Werror -Wformat=2 
-Winit-self -Winline -Wmissing-declarations         
-Wmissing-format-attribute -Wmissing-prototypes         -Wshadow 
-Wsign-compare -Wstrict-prototypes -Wundef -Wunused    -Wwrite-strings 
-fno-strict-aliasing -pipe -I../include 
-DDEFAULT_INCLUDE_PATH=\"/usr/local/etc/nftables\" -g -O2 -MT scanner.o 
-MD -MP -MF $depbase.Tpo -c -o scanner.o scanner.c &&\
        mv -f $depbase.Tpo $depbase.Po
scanner.l:23:20: Fehler: parser.h: Datei oder Verzeichnis nicht gefunden
<stdout>:1395: Fehler: expected specifier-qualifier-list before »YYSTYPE«
<stdout>:1454: Fehler: expected »=«, »,«, »;«, »asm« or »__attribute__« 
before »*« token...

About -Wshadow: in my experience it usually at some point becomes a nuisance
and starts causing strange naming. But for now I'm willing to try, your fix
looks simple enough.


--
To unsubscribe from this list: send the line "unsubscribe netfilter-devel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: nftables-goes-automake patch series

[Jan Engelhardt]

On Wednesday 2009-03-25 14:11, Patrick McHardy wrote:
> Jan Engelhardt wrote:
>> Not much more to elaborate on, this was straight forward to do
>> and only a matter of time, and really needed :)
>
> Patch 1 looks fine, although I might take the liberty to rename the
> file again :)
>
> About automake - I'm glad you're fixing up the makefiles, but I'm
> quite attached to the silent build and the automake version that
> comes with Debian unstable is 1.10, so I assume it won't be
> silent anymore.

I do have a patch for 1.10 too, if you need.

> It also doesn't build anymore on my system, the
> parser.h file is not generated:
>
> make  all-recursive
> make[1]: Entering directory `/tmp/nftables'
> Making all in doc
> make[2]: Entering directory `/tmp/nftables/doc'
> make[2]: Für das Ziel »all« ist nichts zu tun.
> make[2]: Leaving directory `/tmp/nftables/doc'
> Making all in files/nftables
> make[2]: Entering directory `/tmp/nftables/files/nftables'
> make[2]: Für das Ziel »all« ist nichts zu tun.
> make[2]: Leaving directory `/tmp/nftables/files/nftables'
> Making all in src
> make[2]: Entering directory `/tmp/nftables/src'
> depbase=`echo scanner.o | sed 's|[^/]*$|.deps/&|;s|\.o$||'`;\
>       gcc -DHAVE_CONFIG_H -I. -I..    -D_FILE_OFFSET_BITS=64 -D_REENTRANT
> -Wall -Waggregate-return    -Wbad-function-cast -Wcast-align
> -Wdeclaration-after-statement  -Werror -Wformat=2 -Winit-self -Winline
> -Wmissing-declarations         -Wmissing-format-attribute -Wmissing-prototypes
> -Wshadow -Wsign-compare -Wstrict-prototypes -Wundef -Wunused    -Wwrite-strings
> -fno-strict-aliasing -pipe -I../include
> -DDEFAULT_INCLUDE_PATH=\"/usr/local/etc/nftables\" -g -O2 -MT scanner.o -MD -MP
> -MF $depbase.Tpo -c -o scanner.o scanner.c &&\
>       mv -f $depbase.Tpo $depbase.Po
> scanner.l:23:20: Fehler: parser.h: Datei oder Verzeichnis nicht gefunden
> <stdout>:1395: Fehler: expected specifier-qualifier-list before »YYSTYPE«
> <stdout>:1454: Fehler: expected »=«, »,«, »;«, »asm« or »__attribute__« before
> »*« token...

It does do it here:

make[2]: Entering directory `/home/jengelh/Coding/nftables/src'
flex --header-file=scanner.h -t scanner.l >scanner.c
/bin/sh ../ylwrap parser.y y.tab.c parser.c y.tab.h parser.h y.output parser.output -- bison -y  -d --skeleton=../../../../../../../..//home/jengelh/Coding/nftables/src/parser-skeleton.c
updating parser.h
depbase=`echo scanner.o | sed 's|[^/]*$|.deps/&|;s|\.o$||'`;\
        gcc -DHAVE_CONFIG_H -I. -I..    -D_FILE_OFFSET_BITS=64 -D_REENTRANT -Wall -Waggregate-return    -Wbad-function-cast -Wcast-align -Wdeclaration-after-statement  -Werror -Wformat=2 -Winit-self -Winline -Wmissing-declarations  -Wmissing-format-attribute -Wmissing-prototypes         -Wshadow -Wsign-compare -Wstrict-prototypes -Wundef -Wunused    -Wwrite-strings -fno-strict-aliasing -pipe -I../include -DDEFAULT_INCLUDE_PATH=\"/usr/local/etc/nftables\" -g -O2 -MT scanner.o -MD -MP -MF $depbase.Tpo -c -o scanner.o scanner.c &&\
        mv -f $depbase.Tpo $depbase.Po

Maybe you need a distclean? Are you sure you are on commit
v0.01-alpha1-23-g85bf078?

> About -Wshadow: in my experience it usually at some point becomes a nuisance
> and starts causing strange naming. But for now I'm willing to try, your fix
> looks simple enough.
>
Yes, but shadow warns about possibly wrong reuses of variables, i.e.
stuff like

	ctx_t ctx;
	{
		ctx_t ctx;
		memset(&ctx, 0, sizeof(ctx)); // intended to nuke the near ctx
		...
		2-5 lines padding
		...
		send(&ctx); // but wanted to send the far ctx.
	}

You could also move it into a standalone function to avoid this problem ;-)
--
To unsubscribe from this list: send the line "unsubscribe netfilter-devel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: nftables-goes-automake patch series

[Patrick McHardy]

Jan Engelhardt wrote:
> On Wednesday 2009-03-25 14:11, Patrick McHardy wrote:
>> Jan Engelhardt wrote:
>>> Not much more to elaborate on, this was straight forward to do
>>> and only a matter of time, and really needed :)
>> Patch 1 looks fine, although I might take the liberty to rename the
>> file again :)
>>
>> About automake - I'm glad you're fixing up the makefiles, but I'm
>> quite attached to the silent build and the automake version that
>> comes with Debian unstable is 1.10, so I assume it won't be
>> silent anymore.
> 
> I do have a patch for 1.10 too, if you need.

I don't really want to compile my own automake. Can't we just keep
the silent build in combination with automake? I'm fine with doing
the final conversion once that version of automake is commonly
available in distributions.

>> It also doesn't build anymore on my system, the
>> parser.h file is not generated:
>>
>> ...
>> scanner.l:23:20: Fehler: parser.h: Datei oder Verzeichnis nicht gefunden
>> <stdout>:1395: Fehler: expected specifier-qualifier-list before »YYSTYPE«
>> <stdout>:1454: Fehler: expected »=«, »,«, »;«, »asm« or »__attribute__« before
>> »*« token...
> 
> It does do it here:
> 
> make[2]: Entering directory `/home/jengelh/Coding/nftables/src'
> flex --header-file=scanner.h -t scanner.l >scanner.c
> /bin/sh ../ylwrap parser.y y.tab.c parser.c y.tab.h parser.h y.output parser.output -- bison -y  -d --skeleton=../../../../../../../..//home/jengelh/Coding/nftables/src/parser-skeleton.c
> updating parser.h
> depbase=`echo scanner.o | sed 's|[^/]*$|.deps/&|;s|\.o$||'`;\
>         gcc -DHAVE_CONFIG_H -I. -I..    -D_FILE_OFFSET_BITS=64 -D_REENTRANT -Wall -Waggregate-return    -Wbad-function-cast -Wcast-align -Wdeclaration-after-statement  -Werror -Wformat=2 -Winit-self -Winline -Wmissing-declarations  -Wmissing-format-attribute -Wmissing-prototypes         -Wshadow -Wsign-compare -Wstrict-prototypes -Wundef -Wunused    -Wwrite-strings -fno-strict-aliasing -pipe -I../include -DDEFAULT_INCLUDE_PATH=\"/usr/local/etc/nftables\" -g -O2 -MT scanner.o -MD -MP -MF $depbase.Tpo -c -o scanner.o scanner.c &&\
>         mv -f $depbase.Tpo $depbase.Po
> 
> Maybe you need a distclean? Are you sure you are on commit
> v0.01-alpha1-23-g85bf078?

I tried both pulling your tree into mine, as well as a fresh clone of
your tree. Let me know if you need more information.
--
To unsubscribe from this list: send the line "unsubscribe netfilter-devel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: nftables-goes-automake patch series

[Jan Engelhardt]

On Wednesday 2009-03-25 15:33, Patrick McHardy wrote:
>>>
>>> About automake - I'm glad you're fixing up the makefiles, but I'm
>>> quite attached to the silent build and the automake version that
>>> comes with Debian unstable is 1.10, so I assume it won't be
>>> silent anymore.
>>
>> I do have a patch for 1.10 too, if you need.
>
> I don't really want to compile my own automake. Can't we just keep
> the silent build in combination with automake?

I am afraid I am not aware of any Makefile construct that would allow us 
to have such (i.e. silent-by-default, and enabled with V=1) without 
having to override the .c.o rules automake would normally provide.
Overriding $(CC) is not going to work.

> I'm fine with doing
> the final conversion once that version of automake is commonly
> available in distributions.

<placeholder for sarcastic note about age of packages outside sid>

>>> It also doesn't build anymore on my system, the
>>> parser.h file is not generated:
>>>
>>> ...
>>> scanner.l:23:20: Fehler: parser.h: Datei oder Verzeichnis nicht gefunden
>>> <stdout>:1395: Fehler: expected specifier-qualifier-list before »YYSTYPE«
>>> <stdout>:1454: Fehler: expected »=«, »,«, »;«, »asm« or »__attribute__«
>>> before
>>> »*« token...
>>
>> Maybe you need a distclean? Are you sure you are on commit
>> v0.01-alpha1-23-g85bf078?
>
> I tried both pulling your tree into mine, as well as a fresh clone of
> your tree. Let me know if you need more information.
>
Tar up the entire nft tree (i.e. with all the autogenerated code) after 
the compiler errored out and send it over (private, no need to burden 
the list), please.
--
To unsubscribe from this list: send the line "unsubscribe netfilter-devel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: nftables-goes-automake patch series

[Patrick McHardy]

Jan Engelhardt wrote:
> On Wednesday 2009-03-25 15:33, Patrick McHardy wrote:
>>>> About automake - I'm glad you're fixing up the makefiles, but I'm
>>>> quite attached to the silent build and the automake version that
>>>> comes with Debian unstable is 1.10, so I assume it won't be
>>>> silent anymore.
>>> I do have a patch for 1.10 too, if you need.
>> I don't really want to compile my own automake. Can't we just keep
>> the silent build in combination with automake?
> 
> I am afraid I am not aware of any Makefile construct that would allow us 
> to have such (i.e. silent-by-default, and enabled with V=1) without 
> having to override the .c.o rules automake would normally provide.
> Overriding $(CC) is not going to work.

Well, I guess I'll give your patch to automake 1.10 a try.

Re: nftables-goes-automake patch series

[Jan Engelhardt]

On Wednesday 2009-03-25 15:33, Patrick McHardy wrote:
>
>>> It also doesn't build anymore on my system, the
>>> parser.h file is not generated:
>>>
>>> ...
>>> scanner.l:23:20: Fehler: parser.h: Datei oder Verzeichnis nicht gefunden
>>> <stdout>:1395: Fehler: expected specifier-qualifier-list before »YYSTYPE«
>>> <stdout>:1454: Fehler: expected »=«, »,«, »;«, »asm« or »__attribute__«
>>> before
>>> »*« token...
>
> I tried both pulling your tree into mine, as well as a fresh clone of
> your tree. Let me know if you need more information.
>
I rebased and updated the tree now (top: 247566dd) with some info I 
found in the automake manual, see if that works for you.
--
To unsubscribe from this list: send the line "unsubscribe netfilter-devel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html