On 6 March 2013 01:28, Guru Prasad <gurupras@xxxxxxxxxxx> wrote: > What I mean is, every time i start the simulation with a new clock > frequency, I have my cpufreq_init function do the following > int steps; > policy->cpuinfo.max_freq = my_cpufreq_get(policy->cpu); //In KHz > policy->cpuinfo.min_freq = 100 * 1000; //In KHz You don't have to set these, but you just need to pass the right table to cpufreq core and call cpufreq_frequency_table_cpuinfo(). These would be set accordingly by this routine. > policy->cpuinfo.transition_latency = 10 * 1000; //In us > policy->cur = my_cpufreq_get(policy->cpu); //(I'm assuming that > the CPU frequency is always at the maximum before the driver takes > control.) > > > // Get the number of cpufreq steps - Each step is 200MHz > steps = get_cpufreq_steps(policy->cpuinfo.max_freq, > policy->cpuinfo.min_freq); > // Malloc the required space based on number of steps > dyn_cpufreq_table = kmalloc(sizeof(struct > cpufreq_frequency_table) * steps, GFP_KERNEL); > // Create entries in the table > create_freq_table(dyn_cpufreq_table, policy); That's okay, but i didn't get how will you get different set of freqs in your table when you boot up. > However, when I use this code, although my debug statements suggest > that the table is created successfully, sysfs shows no cpufreq under > /sys/devices/system/cpu/cpu0/ Something is wrong. And for finding out what, you need to post your code here, even if it is non-mainline-able. > Am I doing something wrong? Is GFP_KERNEL the wrong flag to use for > this purpose? That's not the culprit for sure. -- To unsubscribe from this list: send the line "unsubscribe cpufreq" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
