Paul Reilly wrote: >>> Can I just copy the whole drivers/char/ipmi/ dir from a 2.6.17 kernel, >>> and put it in my 2.6.16.33 ? >>> >>> >> That is probably the easiest way to handle this. There may be one or >> two things that need to be fixed, but not much. >> > > OK, Thanks Corey. I tried that and it bails out during the kernel compile > with problems with function declarations: > > CC drivers/char/ipmi/ipmi_msghandler.o > drivers/char/ipmi/ipmi_msghandler.c: In function > ?€˜ipmi_init_msghandler?€™: > drivers/char/ipmi/ipmi_msghandler.c:3762: warning: implicit declaration of > function ?€˜atomic_notifier_chain_register?€™ > drivers/char/ipmi/ipmi_msghandler.c: In function > ?€˜cleanup_ipmi?€™: > drivers/char/ipmi/ipmi_msghandler.c:3782: warning: implicit declaration of > function ?€˜atomic_notifier_chain_unregister?€™ > You will need to change the atomic_notifier_chain_[un]register back to notifier_chain_[un]register. The atomic stuff was new in 2.6.17. > CC drivers/char/ipmi/ipmi_devintf.o > CC drivers/char/ipmi/ipmi_si_intf.o > drivers/char/ipmi/ipmi_si_intf.c:240: warning: type defaults to > ?€˜int?€™ in declaration of > ?€˜ATOMIC_NOTIFIER_HEAD?€™ > drivers/char/ipmi/ipmi_si_intf.c:240: warning: parameter names (without types) > in function declaration > drivers/char/ipmi/ipmi_si_intf.c: In function > ?€˜register_xaction_notifier?€™: > drivers/char/ipmi/ipmi_si_intf.c:243: warning: implicit declaration of function > ?€˜atomic_notifier_chain_register?€™ > drivers/char/ipmi/ipmi_si_intf.c:243: error: > ?€˜xaction_notifier_list?€™ undeclared (first use in this > function) > drivers/char/ipmi/ipmi_si_intf.c:243: error: (Each undeclared identifier is > reported only once > drivers/char/ipmi/ipmi_si_intf.c:243: error: for each function it appears in.) > drivers/char/ipmi/ipmi_si_intf.c: In function > ?€˜start_next_msg?€™: > drivers/char/ipmi/ipmi_si_intf.c:305: warning: implicit declaration of function > ?€˜atomic_notifier_call_chain?€™ > drivers/char/ipmi/ipmi_si_intf.c:305: error: > ?€˜xaction_notifier_list?€™ undeclared (first use in this > function) > Like the changes to ipmi_msghandler, the xaction stuff will need to change back to the non-atomic version. Look in the original 2.6.16 code for how it should look. As Phil Pokorny mentioned, IPMI is quite a bit more capable than lmsensors. If you can use IPMI directly, it is very powerful. You have several options of libraries and packages to use it. ipmitool is good for scripting. OpenIPMI is good for writing programs dealing with IPMI. There is some learning curve involved when dealing with IPMI, though. -Corey > make[3]: *** [drivers/char/ipmi/ipmi_si_intf.o] Error 1 > make[2]: *** [drivers/char/ipmi] Error 2 > make[1]: *** [drivers/char] Error 2 > make: *** [drivers] Error 2 > > > This is what I did: > # rm -rf linux-2.6.16.33/drivers/char/ipmi/ > # cp -pR linux-2.6.17.1/drivers/char/ipmi/ linux-2.6.16.33/drivers/char/ > # cp linux-2.6.17.1/include/linux/ipmi* linux-2.6.16.33/include/linux/ > # patch -p1 < /sources/local/bmcsensors/ipmisensors-20060805-0622.diff > patching file drivers/char/ipmi/ipmi_msghandler.c > Hunk #1 succeeded at 1749 (offset 5 lines). > Hunk #2 succeeded at 3839 (offset 8 lines). > patching file drivers/hwmon/Kconfig > Hunk #1 succeeded at 164 (offset -16 lines). > patching file drivers/hwmon/Makefile > Hunk #1 succeeded at 24 with fuzz 1 (offset -2 lines). > patching file drivers/hwmon/ipmisensors.c > patching file drivers/hwmon/ipmisensors.h > patching file include/linux/ipmi.h > Hunk #1 succeeded at 288 (offset 4 lines). > patching file include/linux/ipmi_msgdefs.h > > # cd linux-2.6.16.33 > # make menuconfig > # make > > Should I be doing something different with the include files? > Any tips on how to fix this? > > Paul >
