> From: Matthijs Kooijman [mailto:matthijs@xxxxxxxx] > Sent: Tuesday, October 01, 2013 2:43 PM > > The hardware offers a 4-bit register containing the number of host > channels. However, the values of these register mean 1-16 host channels, > not 0-15. Since the dwc2_hw_params struct stores the actual number of > host channels supported instead of the raw register value, it should be > 5 bits wide instead of 4. > > Before this commit, hardware with 16 host channels would overflow the > field, making it appear as 0 channels. > > This bug was introduced in commit 9badec2 (staging: dwc2: interpret all > hwcfg and related register at init time). > > Reported-by: Dinh Nguyen <dinh.linux@xxxxxxxxx> > Signed-off-by: Matthijs Kooijman <matthijs@xxxxxxxx> > --- > drivers/staging/dwc2/core.h | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/drivers/staging/dwc2/core.h b/drivers/staging/dwc2/core.h > index f7ba34b..fab718d 100644 > --- a/drivers/staging/dwc2/core.h > +++ b/drivers/staging/dwc2/core.h > @@ -294,7 +294,7 @@ struct dwc2_hw_params { > unsigned dev_token_q_depth:5; > unsigned max_transfer_size:26; > unsigned max_packet_count:11; > - unsigned host_channels:4; > + unsigned host_channels:5; > unsigned hs_phy_type:2; > unsigned fs_phy_type:2; > unsigned i2c_enable:1; > -- Acked-by: Paul Zimmerman <paulz@xxxxxxxxxxxx> By the way, it looks like 'num_dev_ep' would have the same problem, except it is not used because we don't support device mode yet. That one and also 'num_dev_perio_in_ep' and 'dev_token_q_depth' should be removed, I think. -- Paul _______________________________________________ devel mailing list devel@xxxxxxxxxxxxxxxxxxxxxx http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel
