2008-02-20 07:19:34 dpmc.S Register mismatch when compiling kernel
Frans Klaver (NETHERLANDS)
Message: 51336 Hi,
When compiling the kernel I run into the following error:
AS arch/blackfin/mach-common/dpmc.o
arch/blackfin/mach-common/dpmc.S: Assembler messages:
arch/blackfin/mach-common/dpmc.S:52: Error: Register mismatch. Input text was ).
make[1]: *** [arch/blackfin/mach-common/dpmc.o] Error 1
make: *** [arch/blackfin/mach-common] Error 2
The line that produces the error in dpmc.S is:
BITSET(R7,(IRQ_WATCH-IVG7));
This apparently is to enable the watchdog timer interrupt. Now after reading a bit of the header files, IRQ_WATCH-IVG7 seems to resolve to 44, thereby trying to set bit 44 of a 32 bit register, which produces this error. As this line of code has existed for over a year, I would think someone would have noticed it, should it be wrong, but I haven't got a clue of whether it's an error in the code or in my configuration.
Disabling power management kind avoids this situation, but doesn't solve it. Power management is most likely going to be an important aspect to me.
I'm building for a BF527-EZkit, with the latest svn sources and using yesterday's automatigically built toolchain.
Thanks in advance,
Frans
config
dpmc.S
QuoteReplyEditDelete
2008-02-20 08:23:59 Re: dpmc.S Register mismatch when compiling kernel
Michael Hennerich (GERMANY)
Message: 51339 you are absolutely right for BF527 this SIC_... Register bit is greater than 32.
My older toolchain for some reason doesn't complain.
This is dead code and should be removed entirely. The only "driver" that uses it is drivers/char/bfin_dpmc. Which is also subject for removal.
The new watchdog driver doesn't rely on this code.
-Michael
QuoteReplyEditDelete
2008-02-20 08:37:34 Re: dpmc.S Register mismatch when compiling kernel
Michael Hennerich (GERMANY)
Message: 51341 Ok I see - the bug only triggers with CONFIG_PM enabled.
Can you do an svn update?
I removed this code.
-Michael
QuoteReplyEditDelete
2008-02-20 09:05:34 Re: dpmc.S Register mismatch when compiling kernel
Frans Klaver (NETHERLANDS)
Message: 51342 Thanks! I'm updating right now. Speaking of which, might there be a chance that my current ip is blacklisted at the svn and who should I contact about that? Currently I have to use a rather inconvenient detour via my computer at home.
QuoteReplyEditDelete
2008-02-20 09:15:11 Re: dpmc.S Register mismatch when compiling kernel
Michael Hennerich (GERMANY)
Message: 51345 Have a look here - maybe this helps:
<a id="using_subversion_sources_behind_a_firewall" name="using_subversion_sources_behind_a_firewall">Using Subversion sources behind a firewall
http://docs.blackfin.uclinux.org/doku.php?id=version_control_systems&s=firewall
-Michael
QuoteReplyEditDelete
2008-02-20 09:26:58 Re: dpmc.S Register mismatch when compiling kernel
Frans Klaver (NETHERLANDS)
Message: 51347 Ah yes, I was afraid you were going to say that. The only option there I haven't yet tried is the rsync one (that one should work though).
QuoteReplyEditDelete
2008-02-20 10:19:40 Re: dpmc.S Register mismatch when compiling kernel
Frans Klaver (NETHERLANDS)
Message: 51352 Kernel build was successful with your fix. Thanks a bunch.
QuoteReplyEditDelete
2008-02-20 12:02:49 Re: dpmc.S Register mismatch when compiling kernel
Mike Frysinger (UNITED STATES)
Message: 51371 we dont blacklist IPs for certain services ... so if you can access the website but not svn, it most likely isnt the server blocking you