> > Jonas Hörberg wrote on 2009-08-10 13:28:27: > > > > > Hi, > > > > The jtag_rclk command should fallback to a fixed frequency for > > interfaces without support for RCLK. > > This is not the present behavior when the command is called in the > > configuration stage (running board config files). > > The reason is that the interface can not be checked for > RCLK support > > before it is initialized. > > The kHz to speed value calculation is also done before the > interface > > is initialized. This will result in the wrong speed setting > when the > > calculation is dependent on data initialized by the init > function of > > the interface. > > Example: ft2232_max_tck for FT2232H and FT4232H based interfaces. > > > > The attached patch contains an attempt to fix these issues > by placing > > the config state speed setting after the initialization of the > > interface. > > > > Best regards > > Jonas Hörberg > > _______________________________________________ > > Openocd-development mailing list > > Openocd-development@lists.berlios.de > > https://lists.berlios.de/mailman/listinfo/openocd-development > > Attached is an new patch that replace the previous one that > had some problems. > * I was not aware of that some interfaces operate on the JTAG > bus in the initialization routine (the j-link issue a TLR). > The clock must be configured by the initialization routine > before the operation in this case. > * The interface initialization functions is expected to configure > the clock speed by using the return value from a call to the > jtag_get_speed function (although this convention is not followed > by some of the interfaces, examples: arm-jtag-ew.c and vsllink.c). > > This patch changes the jtag_get_speed function. Previously it > only returned the value set by the latest call to jtag_set_speed. > Now it calculate the speed value based on the latest speed, > khz or rclk setting. The correct value will now be returned > when it is called from the interface initialization function. > > The name of the jtag_set_speed function is changed to > jtag_config_speed to be aligned with jtag_config_khz and > jtag_config_rclk. > > The global jtag_set_khz function is merged into > jtag_khz_to_speed because external access do not seem to be sane. >
This looks good to me, have only tested the ftdi interface however. No objections/comments before i commit? Cheers Spen _______________________________________________ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
