>
>IMAN_IE is BIT(1), so these macro are respectively equivalent to BIT(1)
>and 0, whatever the value of 'p'.
>
>The purpose was to set and reset a single bit in 'p'.
>Fix these macros to do that correctly.
>
>Fixes: e93e58d27402 ("usb: cdnsp: Device side header file for CDNSP driver")
>Signed-off-by: Christophe JAILLET <christophe.jaillet@xxxxxxxxxx>
Acked-by: Pawel Laszczak <pawell@xxxxxxxxxxx>
Currently the bit 1(RW) and bit 0 (W1toClr) are implemented.
All other bits are not used and are reserved for future RW implementations.
I was a bit afraid about bit 0 which is "write 1 to clear" but I haven't found any
issue with this patch during testing so we can simplify these macros
Thanks Christophe
>---
>This patch is speculative. It is not tested.
>Please, review with care.
>
>Actually, the usage of these macros is:
> readl(somewhere)
> set or reset the IMAN_IE bit
> write(somewhere)
>So it is likely that we want to preserve the other bits read. Otherwise,
>the code could be much simpler.
>---
> drivers/usb/cdns3/cdnsp-gadget.h | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
>diff --git a/drivers/usb/cdns3/cdnsp-gadget.h b/drivers/usb/cdns3/cdnsp-gadget.h
>index 783ca8ffde00..f740fa6089d8 100644
>--- a/drivers/usb/cdns3/cdnsp-gadget.h
>+++ b/drivers/usb/cdns3/cdnsp-gadget.h
>@@ -383,8 +383,8 @@ struct cdnsp_intr_reg {
> #define IMAN_IE BIT(1)
> #define IMAN_IP BIT(0)
> /* bits 2:31 need to be preserved */
>-#define IMAN_IE_SET(p) (((p) & IMAN_IE) | 0x2)
>-#define IMAN_IE_CLEAR(p) (((p) & IMAN_IE) & ~(0x2))
>+#define IMAN_IE_SET(p) ((p) | IMAN_IE)
>+#define IMAN_IE_CLEAR(p) ((p) & ~IMAN_IE)
>
> /* IMOD - Interrupter Moderation Register - irq_control bitmasks. */
> /*
>--
>2.30.2
--
Regards,
Pawel Laszczak
