On 03.08.2014 00:09, Russell King - ARM Linux wrote: > On Thu, Jul 17, 2014 at 06:38:56PM +0200, Tomasz Figa wrote: >> Certain implementations of secure hypervisors (namely the one found on >> Samsung Exynos-based boards) do not provide access to individual L2C >> registers. This makes the .write_sec()-based interface insufficient and >> provoking ugly hacks. >> >> This patch is first step to make the driver not rely on availability of >> writes to individual registers. This is achieved by refactoring the >> driver to use a commit-like operation scheme: all register values are >> prepared first and stored in an instance of l2x0_regs struct and then a >> single callback is responsible to flush those values to the hardware. > > This isn't going to work very well... > >> +static const struct l2c_init_data *l2x0_data; > > So you keep a pointer to the init data... > >> +static void l2c_resume(void) >> +{ >> + l2x0_data->enable(l2x0_base, l2x0_saved_regs.aux_ctrl, >> + l2x0_data->num_lock); > > which you dereference at resume time... > >> static const struct l2c_init_data l2c210_data __initconst = { > > but the structures which get assigned to the pointer are marked __initconst. Good catch. The code was tested on Exynos which requires the cache to be resumed from early assembly code and so I did not hit issues caused by this. Proposed solution: kmemdup() in init, so that only used data remain in memory and the structures can be kept __initconst. Best regards, Tomasz -- To unsubscribe from this list: send the line "unsubscribe linux-omap" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html