[Click] big endian supporting?

Thu Oct 20 11:35:09 EDT 2005

Hi YenJung,

I think you are correct.  I was thinking through this and there are
probably a lot of other small things like this that need fixing.

What kind of board are you running on?

John

YenJung Chang [yenjung at gmail.com] wrote:
>Hi,
>
>I made mistake. I tried to align an aligned configuration. :(.
>
>I think the root cause of this "failed checksum" is:
>There is no byte order conversion to some variables in link_probe, for
>example, _psz in this case of "failed checksum".
>Therefore, the big-endian device would access lp->_psz wrong if the
>value is transmited from a little-endian device, and vice versa.
>
>Is it right?
>
>Thanks,
>YJ.
>
>
>2005/10/5, YenJung Chang <yenjung at gmail.com>:
>> Hi,
>>
>> test.click is attached.(it is same with the roofnet.click I attached
>> in previous mail. :) ).
>>
>> click-align did not add any Align element into test-align.click
>> (test-align.click is same with test.click), and the same message
>> ("ETTStat srcr/es: failed checksum") is printed when I ran "click
>> test-align.click".
>>
>> Thanks,
>> YJ.
>>
>>
>> 2005/10/4, Eddie Kohler <kohler at cs.ucla.edu>:
>> > How did click-align fail?  Can you resend your test.click?
>> >
>> > On Oct 4, 2005, at 5:52 AM, YenJung Chang wrote:
>> >
>> > > Hi,
>> > >
>> > > I guess the root cause is the alignment is needed in my big-endian
>> > > board.
>> > > The problem of ETHTYPE could be fixed if I change the data type of _et
>> > > from uint16_t to uint32_t in ettstat.hh and srforwarder.hh.
>> > > However, another problem followed, it showed "ETTStat srcr/es: failed
>> > > checksum" repeatly.
>> > > I guess this problem is same with ETHTYPE.
>> > >
>> > > I have tried the tool click-align, but failed. (click-align test.click
>> > >
>> > >> test-align.click)
>> > >>
>> > > Is there other way to fix this or have to change all uint16_t to
>> > > uint32_t manually?
>> > >
>> > > Thanks in advance.
>> > > YJ.
>> > >
>> > >
>> > > 2005/10/3, John Bicket <jbicket at amsterdam.lcs.mit.edu>:
>> > >
>> > >> Hi YenJung,
>> > >>
>> > >> You are correct, thanks for the fix; I checked it in.
>> > >>
>> > >>
>> > >> YenJung Chang [yenjung at gmail.com] wrote:
>> > >>
>> > >>> In addition, I could not build click for the big-endian board
>> > >>> successfully before I hacked include/click/glue.hh.
>> > >>>
>> > >>> This is the output message.
>> > >>> --------------------------------------------------------------------
>> > >>> ------------------------------
>> > >>> mips-linux-g++ -W -Wall -fno-exceptions -fno-rtti  -DHAVE_CONFIG_H
>> > >>> -I../include -I../include -I. -I..  -DCLICK_USERLEVEL -static -MD -c
>> > >>> ../elements/wifi/analysis/printpower.cc
>> > >>> ../elements/wifi/analysis/printpower.cc: In constructor `
>> > >>>   PrintPower::PrintPower()':
>> > >>> ../elements/wifi/analysis/printpower.cc:32: warning:
>> > >>> `__base_ctor' is
>> > >>>   deprecated (declared at ../include/click/element.hh:33)
>> > >>> mips-linux-g++ -W -Wall -fno-exceptions -fno-rtti  -DHAVE_CONFIG_H
>> > >>> -I../include -I../include -I. -I..  -DCLICK_USERLEVEL -static -MD -c
>> > >>> ../elements/wifi/analysis/printrate.cc
>> > >>> ../elements/wifi/analysis/printrate.cc: In constructor
>> > >>> `PrintRate::PrintRate()
>> > >>>   ':
>> > >>> ../elements/wifi/analysis/printrate.cc:32: warning: `__base_ctor'
>> > >>> is deprecated
>> > >>>   (declared at ../include/click/element.hh:33)
>> > >>> mips-linux-g++ -W -Wall -fno-exceptions -fno-rtti  -DHAVE_CONFIG_H
>> > >>> -I../include -I../include -I. -I..  -DCLICK_USERLEVEL -static -MD -c
>> > >>> ../elements/wifi/analysis/printseq.cc
>> > >>> ../elements/wifi/analysis/printseq.cc: In member function
>> > >>> `virtual Packet*
>> > >>>   PrintSeq::simple_action(Packet*)':
>> > >>> ../elements/wifi/analysis/printseq.cc:68: error: `bswap_16'
>> > >>> undeclared (first
>> > >>>   use this function)
>> > >>> ../elements/wifi/analysis/printseq.cc:68: error: (Each undeclared
>> > >>> identifier is
>> > >>>   reported only once for each function it appears in.)
>> > >>> ../elements/wifi/analysis/printseq.cc:70: error: `bswap_32'
>> > >>> undeclared (first
>> > >>>   use this function)
>> > >>> make[1]: *** [printseq.o] Error 1
>> > >>> make[1]: Leaving directory `/home1/spencer/roofnet/
>> > >>> click-2005.10.03/userlevel'
>> > >>> make: *** [userlevel] Error 2
>> > >>> --------------------------------------------------------------------
>> > >>> --------------------------------
>> > >>>
>> > >>> This is what I added in include/click/glue.hh
>> > >>> --------------------------------------------------------------------
>> > >>> ----------------------------
>> > >>> Index: include/click/glue.hh
>> > >>> ===================================================================
>> > >>> RCS file: /cvs/click/release/one/include/click/glue.hh,v
>> > >>> retrieving revision 1.50
>> > >>> diff -u -r1.50 glue.hh
>> > >>> --- include/click/glue.hh       25 Aug 2005 04:34:59 -0000      1.50
>> > >>> +++ include/click/glue.hh       3 Oct 2005 06:18:38 -0000
>> > >>> @@ -69,6 +69,10 @@
>> > >>> #  include "simclick.h"
>> > >>> # endif
>> > >>>
>> > >>> +# if CLICK_BYTE_ORDER == CLICK_BIG_ENDIAN
>> > >>> +#  include <byteswap.h>
>> > >>> +# endif
>> > >>> +
>> > >>> #endif
>> > >>> --------------------------------------------------------------------
>> > >>> ----------------------------
>> > >>>
>> > >>> This is the steps I executed.
>> > >>> --------------------------------------------------------------------
>> > >>> ----------------------------
>> > >>> ./configure --enable-wifi --disable-linuxmodule
>> > >>> --build=i686-pc-linux-gnu --host=mipsel-linux --enable-tools=mixed
>> > >>> CLFLAGS=-static CXXFLAGS=-static
>> > >>> make
>> > >>> --------------------------------------------------------------------
>> > >>> ----------------------------
>> > >>>
>> > >>> Please refer to attached config.log for more information.
>> > >>>
>> > >>> After I added those three lines in include/click/glue.hh,
>> > >>> userlevel/click could be built successfully. Is this a bug?
>> > >>> (Although userlevel/click was built, but it could not work well as
>> > >>> described in previous mails.)
>> > >>>
>> > >>> Thanks,
>> > >>> YJ.
>> > >>>
>> > >>> 2005/10/3, YenJung Chang <yenjung at gmail.com>:
>> > >>>
>> > >>>> BTW, I have had the click for x86 use this roofnet.click, it worked
>> > >>>> well; there was no error messages and srcr interface was created
>> > >>>> successfully at least.
>> > >>>>
>> > >>>> Thanks,
>> > >>>> YJ.
>> > >>>>
>> > >>>> 2005/9/30, YenJung Chang <yenjung at gmail.com>:
>> > >>>>
>> > >>>>> Thanks for your reply.
>> > >>>>>
>> > >>>>> I am planing to use Click (userlevel) in a big endian board.
>> > >>>>> And when
>> > >>>>> I run "click roofnet.click", the following message is out.
>> > >>>>> .....
>> > >>>>> /home/am0/biswas/roofnet.click:27: While configuring 'srcr/
>> > >>>>> es :: ETTStat':
>> > >>>>>   Must specify ETHTYPE
>> > >>>>> /home/am0/biswas/roofnet.click:41: While configuring 'srcr/
>> > >>>>> forwarder
>> > >>>>> :: SRForwarder'
>> > >>>>> :
>> > >>>>>   ETHTYPE not specified
>> > >>>>> Router could not be initialized!
>> > >>>>> Segmentation fault
>> > >>>>>
>> > >>>>> Please refer to the attached for the roofnet.click, and tell me
>> > >>>>> if you
>> > >>>>> need more information. :)
>> > >>>>>
>> > >>>>> Thansk.
>> > >>>>> YJ.
>> > >>>>>
>> > >>>>>
>> > >>>>>
>> > >>>>>
>> > >>>>>
>> > >>>>>
>> > >>>>
>> > >>>>
>> > >>
>> > >>
>> > >>
>> > >>> _______________________________________________
>> > >>> click mailing list
>> > >>> click at amsterdam.lcs.mit.edu
>> > >>> https://amsterdam.lcs.mit.edu/mailman/listinfo/click
>> > >>>
>> > >>
>> > >>
>> > >>
>> > >
>> > >
>> >
>> >
>>
>>
>>
>
>_______________________________________________
>click mailing list
>click at amsterdam.lcs.mit.edu
>https://amsterdam.lcs.mit.edu/mailman/listinfo/click