| Message ID | 1415278311-3090-1-git-send-email-smoch@web.de |
|---|---|
| State | Not Applicable |
| Delegated to: | Marek Vasut |
| Headers | show |
On Thursday, November 06, 2014 at 01:51:51 PM, Soeren Moch wrote: > Not only skip storage devices with DEV_TYPE_UNKNOWN, but also all devices > which are unknown to u-boot (e.g., are not HARDDISK, TAPE, CDROM, OPDISK). > > This especially avoids long timeouts when probing for external usb > harddisks which provide "Enclosure Services". > > Signed-off-by: Soeren Moch <smoch@web.de> > -- > Cc: Marek Vasut <marex@denx.de> > Cc: Tom Rini <trini@ti.com> > --- > common/usb_storage.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/common/usb_storage.c b/common/usb_storage.c > index eb7706c..0ac7b48 100644 > --- a/common/usb_storage.c > +++ b/common/usb_storage.c > @@ -1351,7 +1351,7 @@ int usb_stor_get_info(struct usb_device *dev, struct > us_data *ss, perq = usb_stor_buf[0]; > modi = usb_stor_buf[1]; > > - if ((perq & 0x1f) == 0x1f) { > + if ((perq & 0x1f) > DEV_TYPE_OPDISK) { Why can't you just blacklist 0xd instead ? I mean, this patch would do a bulk blacklisting of all the obscure devices with peripheral ID above 0x7, but might still work with this layer (like 0xc ... the RAID controller ; or 0xe ... the reduced block device). Won't it make sense to just selectively blacklist the 0xd ? > /* skip unknown devices */ > return 0; > } Best regards, Marek Vasut
On 11/06/14 21:41, Marek Vasut wrote: > On Thursday, November 06, 2014 at 01:51:51 PM, Soeren Moch wrote: >> Not only skip storage devices with DEV_TYPE_UNKNOWN, but also all devices >> which are unknown to u-boot (e.g., are not HARDDISK, TAPE, CDROM, OPDISK). >> >> This especially avoids long timeouts when probing for external usb >> harddisks which provide "Enclosure Services". >> >> Signed-off-by: Soeren Moch <smoch@web.de> >> -- >> Cc: Marek Vasut <marex@denx.de> >> Cc: Tom Rini <trini@ti.com> >> --- >> common/usb_storage.c | 2 +- >> 1 file changed, 1 insertion(+), 1 deletion(-) >> >> diff --git a/common/usb_storage.c b/common/usb_storage.c >> index eb7706c..0ac7b48 100644 >> --- a/common/usb_storage.c >> +++ b/common/usb_storage.c >> @@ -1351,7 +1351,7 @@ int usb_stor_get_info(struct usb_device *dev, struct >> us_data *ss, perq = usb_stor_buf[0]; >> modi = usb_stor_buf[1]; >> >> - if ((perq & 0x1f) == 0x1f) { >> + if ((perq & 0x1f) > DEV_TYPE_OPDISK) { > > Why can't you just blacklist 0xd instead ? I mean, this patch would do a bulk > blacklisting of all the obscure devices with peripheral ID above 0x7, but might > still work with this layer (like 0xc ... the RAID controller ; or 0xe ... the > reduced block device). > > Won't it make sense to just selectively blacklist the 0xd ? If you think it makes more sense to blacklist 0xd only, that is fine with me. I will prepare a new patch for this. Best regards, Soeren Moch > >> /* skip unknown devices */ >> return 0; >> } > > Best regards, > Marek Vasut >
On Friday, November 07, 2014 at 10:28:18 AM, Soeren Moch wrote: > On 11/06/14 21:41, Marek Vasut wrote: > > On Thursday, November 06, 2014 at 01:51:51 PM, Soeren Moch wrote: > >> Not only skip storage devices with DEV_TYPE_UNKNOWN, but also all > >> devices which are unknown to u-boot (e.g., are not HARDDISK, TAPE, > >> CDROM, OPDISK). > >> > >> This especially avoids long timeouts when probing for external usb > >> harddisks which provide "Enclosure Services". > >> > >> Signed-off-by: Soeren Moch <smoch@web.de> > >> -- > >> Cc: Marek Vasut <marex@denx.de> > >> Cc: Tom Rini <trini@ti.com> > >> --- > >> > >> common/usb_storage.c | 2 +- > >> 1 file changed, 1 insertion(+), 1 deletion(-) > >> > >> diff --git a/common/usb_storage.c b/common/usb_storage.c > >> index eb7706c..0ac7b48 100644 > >> --- a/common/usb_storage.c > >> +++ b/common/usb_storage.c > >> @@ -1351,7 +1351,7 @@ int usb_stor_get_info(struct usb_device *dev, > >> struct us_data *ss, perq = usb_stor_buf[0]; > >> > >> modi = usb_stor_buf[1]; > >> > >> - if ((perq & 0x1f) == 0x1f) { > >> + if ((perq & 0x1f) > DEV_TYPE_OPDISK) { > > > > Why can't you just blacklist 0xd instead ? I mean, this patch would do a > > bulk blacklisting of all the obscure devices with peripheral ID above > > 0x7, but might still work with this layer (like 0xc ... the RAID > > controller ; or 0xe ... the reduced block device). > > > > Won't it make sense to just selectively blacklist the 0xd ? > > If you think it makes more sense to blacklist 0xd only, that is fine with > me. I will prepare a new patch for this. Please also make sure to document why this change is in place. (because the Enclosure Services cause trouble). A simple comment in the code would be nice. Thank you! Best regards, Marek Vasut
diff --git a/common/usb_storage.c b/common/usb_storage.c index eb7706c..0ac7b48 100644 --- a/common/usb_storage.c +++ b/common/usb_storage.c @@ -1351,7 +1351,7 @@ int usb_stor_get_info(struct usb_device *dev, struct us_data *ss, perq = usb_stor_buf[0]; modi = usb_stor_buf[1]; - if ((perq & 0x1f) == 0x1f) { + if ((perq & 0x1f) > DEV_TYPE_OPDISK) { /* skip unknown devices */ return 0; }
Not only skip storage devices with DEV_TYPE_UNKNOWN, but also all devices which are unknown to u-boot (e.g., are not HARDDISK, TAPE, CDROM, OPDISK). This especially avoids long timeouts when probing for external usb harddisks which provide "Enclosure Services". Signed-off-by: Soeren Moch <smoch@web.de> -- Cc: Marek Vasut <marex@denx.de> Cc: Tom Rini <trini@ti.com> --- common/usb_storage.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)