On 05/06/15 16:23, Peter Maydell wrote: > On 26 May 2015 at 17:03, Leon Alrae <leon.al...@imgtec.com> wrote: >> --- a/target-arm/arm-semi.c >> +++ b/target-arm/arm-semi.c >> @@ -27,6 +27,7 @@ >> #include <time.h> >> >> #include "cpu.h" >> +#include "exec/semihost.h" >> #ifdef CONFIG_USER_ONLY >> #include "qemu.h" >> >> @@ -440,10 +441,7 @@ uint32_t do_arm_semihosting(CPUARMState *env) >> input_size = arg1; >> /* Compute the size of the output string. */ >> #if !defined(CONFIG_USER_ONLY) >> - output_size = strlen(ts->boot_info->kernel_filename) >> - + 1 /* Separating space. */ >> - + strlen(ts->boot_info->kernel_cmdline) >> - + 1; /* Terminating null byte. */ >> + output_size = strlen(semihosting_get_cmdline()) + 1; > > It looks like semihosting_get_cmdline() can return NULL, > in which case this will blow up, I think.
semihosting_get_cmdline() returns NULL if neither semihosting args nor -kernel have been specified. As far as I can tell existing implementation may also blow up if kernel_filename is NULL, so we retain the same behaviour. Besides, it's not clear to me how the TARGET_SYS_GET_CMDLINE should behave if cmdline is not available, whether should return -1 or pass an empty string to the guest. For me this looks like a separate issue, not much related to this patch series. Thanks, Leon
