Hi Heinrich, On Thu, 8 Aug 2024 at 15:12, Heinrich Schuchardt <[email protected]> wrote: > > On 07.08.24 16:36, Simon Glass wrote: > > Hi Heinrich, > > > > On Tue, 6 Aug 2024 at 19:56, Heinrich Schuchardt <[email protected]> wrote: > >> > >> On 06.08.24 14:58, Simon Glass wrote: > >>> Set up a function for this, since it needs to be used from multiple test > >>> files. > >>> > >>> Signed-off-by: Simon Glass <[email protected]> > >>> --- > >>> > >>> (no changes since v1) > >>> > >>> test/boot/bootdev.c | 19 ++++++------------- > >>> test/boot/bootstd_common.c | 12 ++++++++++++ > >>> test/boot/bootstd_common.h | 8 ++++++++ > >>> 3 files changed, 26 insertions(+), 13 deletions(-) > >>> > >>> diff --git a/test/boot/bootdev.c b/test/boot/bootdev.c > >>> index 1bf5929c396..de16a51956d 100644 > >>> --- a/test/boot/bootdev.c > >>> +++ b/test/boot/bootdev.c > >>> @@ -16,13 +16,6 @@ > >>> #include <test/ut.h> > >>> #include "bootstd_common.h" > >>> > >>> -/* Allow reseting the USB-started flag */ > >>> -#if defined(CONFIG_USB_HOST) || defined(CONFIG_USB_GADGET) > >>> -extern bool usb_started; > >>> -#else > >>> -#include <usb.h> > >>> -#endif > >>> - > >>> /* Check 'bootdev list' command */ > >>> static int bootdev_test_cmd_list(struct unit_test_state *uts) > >>> { > >>> @@ -201,7 +194,7 @@ static int bootdev_test_order(struct unit_test_state > >>> *uts) > >>> test_set_skip_delays(true); > >>> > >>> /* Start up USB which gives us three additional bootdevs */ > >>> - usb_started = false; > >>> + bootstd_reset_usb(); > >>> ut_assertok(run_command("usb start", 0)); > >>> > >>> /* > >>> @@ -317,7 +310,7 @@ static int bootdev_test_prio(struct unit_test_state > >>> *uts) > >>> test_set_eth_enable(false); > >>> > >>> /* Start up USB which gives us three additional bootdevs */ > >>> - usb_started = false; > >>> + bootstd_reset_usb(); > >>> ut_assertok(run_command("usb start", 0)); > >>> > >>> ut_assertok(bootstd_test_drop_bootdev_order(uts)); > >>> @@ -357,7 +350,7 @@ static int bootdev_test_hunter(struct unit_test_state > >>> *uts) > >>> { > >>> struct bootstd_priv *std; > >>> > >>> - usb_started = false; > >>> + bootstd_reset_usb(); > >>> test_set_skip_delays(true); > >>> > >>> /* get access to the used hunters */ > >>> @@ -398,7 +391,7 @@ static int bootdev_test_cmd_hunt(struct > >>> unit_test_state *uts) > >>> struct bootstd_priv *std; > >>> > >>> test_set_skip_delays(true); > >>> - usb_started = false; > >>> + bootstd_reset_usb(); > >>> > >>> /* get access to the used hunters */ > >>> ut_assertok(bootstd_get_priv(&std)); > >>> @@ -527,7 +520,7 @@ BOOTSTD_TEST(bootdev_test_bootable, UT_TESTF_DM | > >>> UT_TESTF_SCAN_FDT); > >>> /* Check hunting for bootdev of a particular priority */ > >>> static int bootdev_test_hunt_prio(struct unit_test_state *uts) > >>> { > >>> - usb_started = false; > >>> + bootstd_reset_usb(); > >>> test_set_skip_delays(true); > >>> > >>> console_record_reset_enable(); > >>> @@ -556,7 +549,7 @@ static int bootdev_test_hunt_label(struct > >>> unit_test_state *uts) > >>> struct bootstd_priv *std; > >>> int mflags; > >>> > >>> - usb_started = false; > >>> + bootstd_reset_usb(); > >>> > >>> /* get access to the used hunters */ > >>> ut_assertok(bootstd_get_priv(&std)); > >>> diff --git a/test/boot/bootstd_common.c b/test/boot/bootstd_common.c > >>> index e50539500a0..ff0aff4bbe7 100644 > >>> --- a/test/boot/bootstd_common.c > >>> +++ b/test/boot/bootstd_common.c > >>> @@ -20,6 +20,13 @@ > >>> /* tracks whether bootstd_setup_for_tests() has been run yet */ > >>> bool vbe_setup_done; > >>> > >>> +/* Allow resetting the USB-started flag */ > >>> +#if defined(CONFIG_USB_HOST) || defined(CONFIG_USB_GADGET) > >>> +extern bool usb_started; > >>> +#else > >>> +#include <usb.h> > >>> +#endif > >> > >> Why can't you always include usb.h which defines usb_started? > >> Is there a bug in usb.h? > > > > I can, but even in tests it is not great to access globals. Having it > > in a function lets us call out these particular settings better, which > > has aided debugging, at least for me. These tests are special in that > > they are not DM tests, but do stop USB. > > I was just wondering if we can can rid of the #if here. That should not > stop you from creating a function to access the variable.
Ah, I see. Yes, it is an artifact of it being built for boards without USB, but now we only build for sandbox, so I can drop that. Regards, Simon
