On Wed, 13 Aug 2008 15:48:14 +0200, Wolfgang Grandegger wrote: > Jean Delvare wrote: > > Your patch breaks the build with CONFIG_ISA=y: > > > > CC [M] drivers/hwmon/w83781d.o > > drivers/hwmon/w83781d.c: In function `w83781d_read_value_i2c_isa': > > drivers/hwmon/w83781d.c:1668: error: `client' undeclared (first use in this function) > > drivers/hwmon/w83781d.c:1668: error: (Each undeclared identifier is reported only once > > drivers/hwmon/w83781d.c:1668: error: for each function it appears in.) > > drivers/hwmon/w83781d.c:1660: warning: unused variable `bank' > > drivers/hwmon/w83781d.c: In function `w83781d_write_value_i2c_isa': > > drivers/hwmon/w83781d.c:1694: warning: unused variable `bank' > > drivers/hwmon/w83781d.c:1695: warning: unused variable `cl' > > drivers/hwmon/w83781d.c: In function `w83781d_isa_add_driver': > > drivers/hwmon/w83781d.c:1967: error: `w83781d_read_value_isa' undeclared (first use in this function) > > drivers/hwmon/w83781d.c:1968: error: `w83781d_write_value_isa' undeclared (first use in this function) > > drivers/hwmon/w83781d.c: At top level: > > drivers/hwmon/w83781d.c:1659: warning: 'w83781d_read_value_i2c_isa' defined but not used > > drivers/hwmon/w83781d.c:1692: warning: 'w83781d_write_value_i2c_isa' defined but not used > > make[2]: *** [drivers/hwmon/w83781d.o] Error 1 > > make[1]: *** [drivers/hwmon] Error 2 > > make: *** [drivers] Error 2 > > > Apparently you didn't test that case - you really have to. > > I thought I did but... Anyway, I cannot really test the ISA interface > because I just have an embedded PowerPC board with a w83782. I can do the testing, no problem with that. Please just make sure that it builds OK. > > (...) > > I don't think that going to this level of optimization is needed. Or, > > if you insist on doing it, that would be a separate patch. The first > > patch should really only move things around to group them inside the > > #ifdef/#endif, and skip the detection of the ISA device and > > registration of the platform driver. You shouldn't have to modify the > > actual driver code. > > I did not do any optimization saving code. I was referring to the split of the register access functions. -- Jean Delvare
