On Tue, May 03, 2022 at 05:27:08PM +0200, Geert Uytterhoeven wrote:
> Hi Alan,
>
> On Tue, May 3, 2022 at 5:14 PM Alan Stern <[email protected]> wrote:
> > On Tue, May 03, 2022 at 12:14:30PM +0200, Geert Uytterhoeven wrote:
> > > On Sat, 23 Apr 2022, Alan Stern wrote:
> > > > This patch adds a "gadget" bus and uses it for registering gadgets and
> > > > their drivers. From now on, bindings will be managed by the driver
> > > > core rather than through ad-hoc manipulations in the UDC core.
> > > >
> > > > As part of this change, the driver_pending_list is removed. The UDC
> > > > core won't need to keep track of unbound drivers for later binding,
> > > > because the driver core handles all of that for us.
> > > >
> > > > However, we do need one new feature: a way to prevent gadget drivers
> > > > from being bound to more than one gadget at a time. The existing code
> > > > does this automatically, but the driver core doesn't -- it's perfectly
> > > > happy to bind a single driver to all the matching devices on the bus.
> > > > The patch adds a new bitflag to the usb_gadget_driver structure for
> > > > this purpose.
> > > >
> > > > A nice side effect of this change is a reduction in the total lines of
> > > > code, since now the driver core will do part of the work that the UDC
> > > > used to do.
> > > >
> > > > A possible future patch could add udc devices to the gadget bus, say
> > > > as a separate device type.
> > > >
> > > > Signed-off-by: Alan Stern <[email protected]>
> > >
> > > Thanks for your patch, which is now commit fc274c1e997314bf ("USB:
> > > gadget: Add a new bus for gadgets") in usb-next.
> > >
> > > This patch cause a regression on the Renesas Salvator-XS development
> > > board, as R-Car H3 has multiple USB gadget devices:
> >
> > Then these gadgets ought to have distinct names in order to avoid the
> > conflict below:
> >
> > > sysfs: cannot create duplicate filename '/bus/gadget/devices/gadget'
> > > CPU: 2 PID: 1 Comm: swapper/0 Not tainted 5.18.0-rc1-arm64-renesas-00074-gfc274c1e9973 #1587
> > > Hardware name: Renesas Salvator-X 2nd version board based on r8a77951 (DT)
> > > Call trace:
> > > dump_backtrace+0xcc/0xd8
> > > show_stack+0x14/0x30
> > > dump_stack_lvl+0x88/0xb0
> > > dump_stack+0x14/0x2c
> > > sysfs_warn_dup+0x60/0x78
> > > sysfs_do_create_link_sd.isra.0+0xe4/0xf0
> > > sysfs_create_link+0x20/0x40
> > > bus_add_device+0x64/0x110
> > > device_add+0x31c/0x850
> > > usb_add_gadget+0x124/0x1a0
> > > usb_add_gadget_udc_release+0x1c/0x50
> > > usb_add_gadget_udc+0x10/0x18
> > > renesas_usb3_probe+0x450/0x728
> > ...
> >
> > Having three gadget devices, all named "gadget", doesn't seem like a
> > good idea.
>
> I'm not so sure where these names are coming from.
> `git grep '"gadget"'` points to the following likely targets:
>
> drivers/usb/gadget/udc/core.c: dev_set_name(&gadget->dev, "gadget");
> drivers/usb/renesas_usbhs/mod_gadget.c: gpriv->mod.name = "gadget";
>
> Changing both names reveals the problem is actually caused by
> the former ;-)
Ah, good.
One way to attack this would be to keep a static counter and dynamically
set the name to "gadget.%d" using the counter's value. Or keep a bitmap
of allocated gadget numbers and use the first available number.
Felipe, Greg, any opinions?
Ironically, the UDC driver itself provides a name in gadget->name. But
that string isn't unique either (in renesas-usb3, for instance, it is
always set to "renesas_usb3"), so it won't help solve this problem.
> > This doesn't seem like it should be too hard to fix, although I'm not
> > at all familiar with the renesas-usb3 driver. Do you know who maintains
> > that driver? Is it you?
>
> Adding Shimoda-san to CC (but he's enjoying Golden Week).
It looks like the problem has to be solved in the gadget core rather
than in the UDC driver. So we won't need to modify the driver after
all.
Alan Stern
On Tue, May 03, 2022 at 11:48:33AM -0400, Alan Stern wrote:
> On Tue, May 03, 2022 at 05:27:08PM +0200, Geert Uytterhoeven wrote:
> > Hi Alan,
> >
> > On Tue, May 3, 2022 at 5:14 PM Alan Stern <[email protected]> wrote:
> > > On Tue, May 03, 2022 at 12:14:30PM +0200, Geert Uytterhoeven wrote:
> > > > On Sat, 23 Apr 2022, Alan Stern wrote:
> > > > > This patch adds a "gadget" bus and uses it for registering gadgets and
> > > > > their drivers. From now on, bindings will be managed by the driver
> > > > > core rather than through ad-hoc manipulations in the UDC core.
> > > > >
> > > > > As part of this change, the driver_pending_list is removed. The UDC
> > > > > core won't need to keep track of unbound drivers for later binding,
> > > > > because the driver core handles all of that for us.
> > > > >
> > > > > However, we do need one new feature: a way to prevent gadget drivers
> > > > > from being bound to more than one gadget at a time. The existing code
> > > > > does this automatically, but the driver core doesn't -- it's perfectly
> > > > > happy to bind a single driver to all the matching devices on the bus.
> > > > > The patch adds a new bitflag to the usb_gadget_driver structure for
> > > > > this purpose.
> > > > >
> > > > > A nice side effect of this change is a reduction in the total lines of
> > > > > code, since now the driver core will do part of the work that the UDC
> > > > > used to do.
> > > > >
> > > > > A possible future patch could add udc devices to the gadget bus, say
> > > > > as a separate device type.
> > > > >
> > > > > Signed-off-by: Alan Stern <[email protected]>
> > > >
> > > > Thanks for your patch, which is now commit fc274c1e997314bf ("USB:
> > > > gadget: Add a new bus for gadgets") in usb-next.
> > > >
> > > > This patch cause a regression on the Renesas Salvator-XS development
> > > > board, as R-Car H3 has multiple USB gadget devices:
> > >
> > > Then these gadgets ought to have distinct names in order to avoid the
> > > conflict below:
> > >
> > > > sysfs: cannot create duplicate filename '/bus/gadget/devices/gadget'
> > > > CPU: 2 PID: 1 Comm: swapper/0 Not tainted 5.18.0-rc1-arm64-renesas-00074-gfc274c1e9973 #1587
> > > > Hardware name: Renesas Salvator-X 2nd version board based on r8a77951 (DT)
> > > > Call trace:
> > > > dump_backtrace+0xcc/0xd8
> > > > show_stack+0x14/0x30
> > > > dump_stack_lvl+0x88/0xb0
> > > > dump_stack+0x14/0x2c
> > > > sysfs_warn_dup+0x60/0x78
> > > > sysfs_do_create_link_sd.isra.0+0xe4/0xf0
> > > > sysfs_create_link+0x20/0x40
> > > > bus_add_device+0x64/0x110
> > > > device_add+0x31c/0x850
> > > > usb_add_gadget+0x124/0x1a0
> > > > usb_add_gadget_udc_release+0x1c/0x50
> > > > usb_add_gadget_udc+0x10/0x18
> > > > renesas_usb3_probe+0x450/0x728
> > > ...
> > >
> > > Having three gadget devices, all named "gadget", doesn't seem like a
> > > good idea.
> >
> > I'm not so sure where these names are coming from.
> > `git grep '"gadget"'` points to the following likely targets:
> >
> > drivers/usb/gadget/udc/core.c: dev_set_name(&gadget->dev, "gadget");
> > drivers/usb/renesas_usbhs/mod_gadget.c: gpriv->mod.name = "gadget";
> >
> > Changing both names reveals the problem is actually caused by
> > the former ;-)
>
> Ah, good.
>
> One way to attack this would be to keep a static counter and dynamically
> set the name to "gadget.%d" using the counter's value. Or keep a bitmap
> of allocated gadget numbers and use the first available number.
>
> Felipe, Greg, any opinions?
Just use an idr structure for the number, that's the simplest way to
track that.
thanks,
greg k-h
On Tue, May 03, 2022 at 11:48:33AM -0400, Alan Stern wrote:
> On Tue, May 03, 2022 at 05:27:08PM +0200, Geert Uytterhoeven wrote:
> > Hi Alan,
> >
> > On Tue, May 3, 2022 at 5:14 PM Alan Stern <[email protected]> wrote:
> > > On Tue, May 03, 2022 at 12:14:30PM +0200, Geert Uytterhoeven wrote:
> > > > On Sat, 23 Apr 2022, Alan Stern wrote:
> > > > > This patch adds a "gadget" bus and uses it for registering gadgets and
> > > > > their drivers. From now on, bindings will be managed by the driver
> > > > > core rather than through ad-hoc manipulations in the UDC core.
> > > > >
> > > > > As part of this change, the driver_pending_list is removed. The UDC
> > > > > core won't need to keep track of unbound drivers for later binding,
> > > > > because the driver core handles all of that for us.
> > > > >
> > > > > However, we do need one new feature: a way to prevent gadget drivers
> > > > > from being bound to more than one gadget at a time. The existing code
> > > > > does this automatically, but the driver core doesn't -- it's perfectly
> > > > > happy to bind a single driver to all the matching devices on the bus.
> > > > > The patch adds a new bitflag to the usb_gadget_driver structure for
> > > > > this purpose.
> > > > >
> > > > > A nice side effect of this change is a reduction in the total lines of
> > > > > code, since now the driver core will do part of the work that the UDC
> > > > > used to do.
> > > > >
> > > > > A possible future patch could add udc devices to the gadget bus, say
> > > > > as a separate device type.
> > > > >
> > > > > Signed-off-by: Alan Stern <[email protected]>
> > > >
> > > > Thanks for your patch, which is now commit fc274c1e997314bf ("USB:
> > > > gadget: Add a new bus for gadgets") in usb-next.
> > > >
> > > > This patch cause a regression on the Renesas Salvator-XS development
> > > > board, as R-Car H3 has multiple USB gadget devices:
> > >
> > > Then these gadgets ought to have distinct names in order to avoid the
> > > conflict below:
Geert:
Can you test the patch below? It ought to fix the problem (although it
might end up causing other problems down the line...)
Alan Stern
Index: usb-devel/drivers/usb/gadget/udc/core.c
===================================================================
--- usb-devel.orig/drivers/usb/gadget/udc/core.c
+++ usb-devel/drivers/usb/gadget/udc/core.c
@@ -12,6 +12,7 @@
#include <linux/module.h>
#include <linux/device.h>
#include <linux/list.h>
+#include <linux/idr.h>
#include <linux/err.h>
#include <linux/dma-mapping.h>
#include <linux/sched/task_stack.h>
@@ -23,6 +24,8 @@
#include "trace.h"
+static DEFINE_IDA(gadget_id_numbers);
+
static struct bus_type gadget_bus_type;
/**
@@ -1248,7 +1251,6 @@ static void usb_udc_nop_release(struct d
void usb_initialize_gadget(struct device *parent, struct usb_gadget *gadget,
void (*release)(struct device *dev))
{
- dev_set_name(&gadget->dev, "gadget");
INIT_WORK(&gadget->work, usb_gadget_state_work);
gadget->dev.parent = parent;
@@ -1304,12 +1306,21 @@ int usb_add_gadget(struct usb_gadget *ga
usb_gadget_set_state(gadget, USB_STATE_NOTATTACHED);
udc->vbus = true;
+ ret = ida_alloc(&gadget_id_numbers, GFP_KERNEL);
+ if (ret < 0)
+ goto err_del_udc;
+ gadget->id_number = ret;
+ dev_set_name(&gadget->dev, "gadget.%d", ret);
+
ret = device_add(&gadget->dev);
if (ret)
- goto err_del_udc;
+ goto err_free_id;
return 0;
+ err_free_id:
+ ida_free(&gadget_id_numbers, gadget->id_number);
+
err_del_udc:
flush_work(&gadget->work);
device_del(&udc->dev);
@@ -1417,6 +1428,7 @@ void usb_del_gadget(struct usb_gadget *g
kobject_uevent(&udc->dev.kobj, KOBJ_REMOVE);
flush_work(&gadget->work);
device_del(&gadget->dev);
+ ida_free(&gadget_id_numbers, gadget->id_number);
device_unregister(&udc->dev);
}
EXPORT_SYMBOL_GPL(usb_del_gadget);
Index: usb-devel/include/linux/usb/gadget.h
===================================================================
--- usb-devel.orig/include/linux/usb/gadget.h
+++ usb-devel/include/linux/usb/gadget.h
@@ -386,6 +386,7 @@ struct usb_gadget_ops {
* @lpm_capable: If the gadget max_speed is FULL or HIGH, this flag
* indicates that it supports LPM as per the LPM ECN & errata.
* @irq: the interrupt number for device controller.
+ * @id_number: a unique ID number for ensuring that gadget names are distinct
*
* Gadgets have a mostly-portable "gadget driver" implementing device
* functions, handling all usb configurations and interfaces. Gadget
@@ -446,6 +447,7 @@ struct usb_gadget {
unsigned connected:1;
unsigned lpm_capable:1;
int irq;
+ int id_number;
};
#define work_to_gadget(w) (container_of((w), struct usb_gadget, work))
Hi Alan,
On Sat, May 7, 2022 at 5:36 PM Alan Stern <[email protected]> wrote:
> On Tue, May 03, 2022 at 11:48:33AM -0400, Alan Stern wrote:
> > On Tue, May 03, 2022 at 05:27:08PM +0200, Geert Uytterhoeven wrote:
> > > On Tue, May 3, 2022 at 5:14 PM Alan Stern <[email protected]> wrote:
> > > > On Tue, May 03, 2022 at 12:14:30PM +0200, Geert Uytterhoeven wrote:
> > > > > On Sat, 23 Apr 2022, Alan Stern wrote:
> > > > > > This patch adds a "gadget" bus and uses it for registering gadgets and
> > > > > > their drivers. From now on, bindings will be managed by the driver
> > > > > > core rather than through ad-hoc manipulations in the UDC core.
> > > > > >
> > > > > > As part of this change, the driver_pending_list is removed. The UDC
> > > > > > core won't need to keep track of unbound drivers for later binding,
> > > > > > because the driver core handles all of that for us.
> > > > > >
> > > > > > However, we do need one new feature: a way to prevent gadget drivers
> > > > > > from being bound to more than one gadget at a time. The existing code
> > > > > > does this automatically, but the driver core doesn't -- it's perfectly
> > > > > > happy to bind a single driver to all the matching devices on the bus.
> > > > > > The patch adds a new bitflag to the usb_gadget_driver structure for
> > > > > > this purpose.
> > > > > >
> > > > > > A nice side effect of this change is a reduction in the total lines of
> > > > > > code, since now the driver core will do part of the work that the UDC
> > > > > > used to do.
> > > > > >
> > > > > > A possible future patch could add udc devices to the gadget bus, say
> > > > > > as a separate device type.
> > > > > >
> > > > > > Signed-off-by: Alan Stern <[email protected]>
> > > > >
> > > > > Thanks for your patch, which is now commit fc274c1e997314bf ("USB:
> > > > > gadget: Add a new bus for gadgets") in usb-next.
> > > > >
> > > > > This patch cause a regression on the Renesas Salvator-XS development
> > > > > board, as R-Car H3 has multiple USB gadget devices:
> > > >
> > > > Then these gadgets ought to have distinct names in order to avoid the
> > > > conflict below:
>
> Geert:
>
> Can you test the patch below? It ought to fix the problem (although it
Thanks!
root@h3-salvator-xs:~# ls -l /sys/bus/gadget/devices/
total 0
lrwxrwxrwx 1 root root 0 Feb 14 2019 gadget.0 ->
../../../devices/platform/soc/e659c000.usb/gadget.0
lrwxrwxrwx 1 root root 0 Feb 14 2019 gadget.1 ->
../../../devices/platform/soc/ee020000.usb/gadget.1
lrwxrwxrwx 1 root root 0 Feb 14 2019 gadget.2 ->
../../../devices/platform/soc/e6590000.usb/gadget.2
Tested-by: Geert Uytterhoeven <[email protected]>
LGTM, so
Reviewed-by: Geert Uytterhoeven <[email protected]>
> might end up causing other problems down the line...)
Can you please elaborate? I'm not too familiar with UBS gadgets.
Gr{oetje,eeting}s,
Geert
--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]
In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds
On Mon, May 09, 2022 at 09:46:25AM +0200, Geert Uytterhoeven wrote:
> > Geert:
> >
> > Can you test the patch below? It ought to fix the problem (although it
>
> Thanks!
>
> root@h3-salvator-xs:~# ls -l /sys/bus/gadget/devices/
> total 0
> lrwxrwxrwx 1 root root 0 Feb 14 2019 gadget.0 ->
> ../../../devices/platform/soc/e659c000.usb/gadget.0
> lrwxrwxrwx 1 root root 0 Feb 14 2019 gadget.1 ->
> ../../../devices/platform/soc/ee020000.usb/gadget.1
> lrwxrwxrwx 1 root root 0 Feb 14 2019 gadget.2 ->
> ../../../devices/platform/soc/e6590000.usb/gadget.2
>
> Tested-by: Geert Uytterhoeven <[email protected]>
>
> LGTM, so
> Reviewed-by: Geert Uytterhoeven <[email protected]>
Thanks!
> > might end up causing other problems down the line...)
>
> Can you please elaborate? I'm not too familiar with UBS gadgets.
I was concerned about the fact that changing the name of a file,
directory, or symbolic link in sysfs means changing a user API, and so
it might cause some existing programs to fail. That would be a
regression.
Perhaps the best way to work around the problem is to leave the name set
to "gadget" if the ID number is 0, while adding the ID number on to the
name if the value is > 0. What do you think?
Alan Stern
Hi Alan,
On Mon, May 9, 2022 at 4:15 PM Alan Stern <[email protected]> wrote:
> On Mon, May 09, 2022 at 09:46:25AM +0200, Geert Uytterhoeven wrote:
> > > Geert:
> > >
> > > Can you test the patch below? It ought to fix the problem (although it
> >
> > Thanks!
> >
> > root@h3-salvator-xs:~# ls -l /sys/bus/gadget/devices/
> > total 0
> > lrwxrwxrwx 1 root root 0 Feb 14 2019 gadget.0 ->
> > ../../../devices/platform/soc/e659c000.usb/gadget.0
> > lrwxrwxrwx 1 root root 0 Feb 14 2019 gadget.1 ->
> > ../../../devices/platform/soc/ee020000.usb/gadget.1
> > lrwxrwxrwx 1 root root 0 Feb 14 2019 gadget.2 ->
> > ../../../devices/platform/soc/e6590000.usb/gadget.2
> >
> > Tested-by: Geert Uytterhoeven <[email protected]>
> >
> > LGTM, so
> > Reviewed-by: Geert Uytterhoeven <[email protected]>
>
> Thanks!
>
> > > might end up causing other problems down the line...)
> >
> > Can you please elaborate? I'm not too familiar with UBS gadgets.
>
> I was concerned about the fact that changing the name of a file,
> directory, or symbolic link in sysfs means changing a user API, and so
> it might cause some existing programs to fail. That would be a
> regression.
>
> Perhaps the best way to work around the problem is to leave the name set
> to "gadget" if the ID number is 0, while adding the ID number on to the
> name if the value is > 0. What do you think?
Oh, you mean the "gadget.N" subdirs, which are the targets of the
symlinks above? These were indeed named "gadget" before.
Would it be possible to append the ".N" suffixes only to the actual
symlinks, while keeping the target directory names unchanged?
E.g. /sys/bus/gadget/devices/gadget.0 ->
../../../devices/platform/soc/e659c000.usb/gadget
Gr{oetje,eeting}s,
Geert
--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]
In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds
On Mon, May 09, 2022 at 04:42:01PM +0200, Geert Uytterhoeven wrote:
> Hi Alan,
>
> On Mon, May 9, 2022 at 4:15 PM Alan Stern <[email protected]> wrote:
> > On Mon, May 09, 2022 at 09:46:25AM +0200, Geert Uytterhoeven wrote:
> > > > Geert:
> > > >
> > > > Can you test the patch below? It ought to fix the problem (although it
> > >
> > > Thanks!
> > >
> > > root@h3-salvator-xs:~# ls -l /sys/bus/gadget/devices/
> > > total 0
> > > lrwxrwxrwx 1 root root 0 Feb 14 2019 gadget.0 ->
> > > ../../../devices/platform/soc/e659c000.usb/gadget.0
> > > lrwxrwxrwx 1 root root 0 Feb 14 2019 gadget.1 ->
> > > ../../../devices/platform/soc/ee020000.usb/gadget.1
> > > lrwxrwxrwx 1 root root 0 Feb 14 2019 gadget.2 ->
> > > ../../../devices/platform/soc/e6590000.usb/gadget.2
> > >
> > > Tested-by: Geert Uytterhoeven <[email protected]>
> > >
> > > LGTM, so
> > > Reviewed-by: Geert Uytterhoeven <[email protected]>
> >
> > Thanks!
> >
> > > > might end up causing other problems down the line...)
> > >
> > > Can you please elaborate? I'm not too familiar with UBS gadgets.
> >
> > I was concerned about the fact that changing the name of a file,
> > directory, or symbolic link in sysfs means changing a user API, and so
> > it might cause some existing programs to fail. That would be a
> > regression.
> >
> > Perhaps the best way to work around the problem is to leave the name set
> > to "gadget" if the ID number is 0, while adding the ID number on to the
> > name if the value is > 0. What do you think?
>
> Oh, you mean the "gadget.N" subdirs, which are the targets of the
> symlinks above? These were indeed named "gadget" before.
>
> Would it be possible to append the ".N" suffixes only to the actual
> symlinks, while keeping the target directory names unchanged?
> E.g. /sys/bus/gadget/devices/gadget.0 ->
> ../../../devices/platform/soc/e659c000.usb/gadget
No, it's not possible. Or at least, not without significant changes to
the driver core. Besides, people expect these names to be the same.
Alan Stern
On Mon, May 09, 2022 at 11:05:06AM -0400, Alan Stern wrote:
> On Mon, May 09, 2022 at 04:42:01PM +0200, Geert Uytterhoeven wrote:
> > Hi Alan,
> >
> > On Mon, May 9, 2022 at 4:15 PM Alan Stern <[email protected]> wrote:
> > > On Mon, May 09, 2022 at 09:46:25AM +0200, Geert Uytterhoeven wrote:
> > > > > Geert:
> > > > >
> > > > > Can you test the patch below? It ought to fix the problem (although it
> > > >
> > > > Thanks!
> > > >
> > > > root@h3-salvator-xs:~# ls -l /sys/bus/gadget/devices/
> > > > total 0
> > > > lrwxrwxrwx 1 root root 0 Feb 14 2019 gadget.0 ->
> > > > ../../../devices/platform/soc/e659c000.usb/gadget.0
> > > > lrwxrwxrwx 1 root root 0 Feb 14 2019 gadget.1 ->
> > > > ../../../devices/platform/soc/ee020000.usb/gadget.1
> > > > lrwxrwxrwx 1 root root 0 Feb 14 2019 gadget.2 ->
> > > > ../../../devices/platform/soc/e6590000.usb/gadget.2
> > > >
> > > > Tested-by: Geert Uytterhoeven <[email protected]>
> > > >
> > > > LGTM, so
> > > > Reviewed-by: Geert Uytterhoeven <[email protected]>
> > >
> > > Thanks!
> > >
> > > > > might end up causing other problems down the line...)
> > > >
> > > > Can you please elaborate? I'm not too familiar with UBS gadgets.
> > >
> > > I was concerned about the fact that changing the name of a file,
> > > directory, or symbolic link in sysfs means changing a user API, and so
> > > it might cause some existing programs to fail. That would be a
> > > regression.
> > >
> > > Perhaps the best way to work around the problem is to leave the name set
> > > to "gadget" if the ID number is 0, while adding the ID number on to the
> > > name if the value is > 0. What do you think?
> >
> > Oh, you mean the "gadget.N" subdirs, which are the targets of the
> > symlinks above? These were indeed named "gadget" before.
> >
> > Would it be possible to append the ".N" suffixes only to the actual
> > symlinks, while keeping the target directory names unchanged?
> > E.g. /sys/bus/gadget/devices/gadget.0 ->
> > ../../../devices/platform/soc/e659c000.usb/gadget
>
> No, it's not possible. Or at least, not without significant changes to
> the driver core. Besides, people expect these names to be the same.
It should always be ok to change the names of devices as those are not
going to be persistent / determinisitic. It's the attributes of devices
that are supposed to be used to determine those types of things.
So I think let's start out with the .N suffix for everything for now.
I'll be glad to submit the fixed patch to the Android kernel build
system to see what their testing comes back with to see if they happened
to make any name assumptions as that's the largest user of this
codebase.
thanks,
greg k-h
On Mon, May 09, 2022 at 06:23:31PM +0200, Greg KH wrote:
> On Mon, May 09, 2022 at 11:05:06AM -0400, Alan Stern wrote:
> > On Mon, May 09, 2022 at 04:42:01PM +0200, Geert Uytterhoeven wrote:
> > > Hi Alan,
> > >
> > > On Mon, May 9, 2022 at 4:15 PM Alan Stern <[email protected]> wrote:
> > > > On Mon, May 09, 2022 at 09:46:25AM +0200, Geert Uytterhoeven wrote:
> > > > > > Geert:
> > > > > >
> > > > > > Can you test the patch below? It ought to fix the problem (although it
> > > > >
> > > > > Thanks!
> > > > >
> > > > > root@h3-salvator-xs:~# ls -l /sys/bus/gadget/devices/
> > > > > total 0
> > > > > lrwxrwxrwx 1 root root 0 Feb 14 2019 gadget.0 ->
> > > > > ../../../devices/platform/soc/e659c000.usb/gadget.0
> > > > > lrwxrwxrwx 1 root root 0 Feb 14 2019 gadget.1 ->
> > > > > ../../../devices/platform/soc/ee020000.usb/gadget.1
> > > > > lrwxrwxrwx 1 root root 0 Feb 14 2019 gadget.2 ->
> > > > > ../../../devices/platform/soc/e6590000.usb/gadget.2
> > > > >
> > > > > Tested-by: Geert Uytterhoeven <[email protected]>
> > > > >
> > > > > LGTM, so
> > > > > Reviewed-by: Geert Uytterhoeven <[email protected]>
> > > >
> > > > Thanks!
> > > >
> > > > > > might end up causing other problems down the line...)
> > > > >
> > > > > Can you please elaborate? I'm not too familiar with UBS gadgets.
> > > >
> > > > I was concerned about the fact that changing the name of a file,
> > > > directory, or symbolic link in sysfs means changing a user API, and so
> > > > it might cause some existing programs to fail. That would be a
> > > > regression.
> > > >
> > > > Perhaps the best way to work around the problem is to leave the name set
> > > > to "gadget" if the ID number is 0, while adding the ID number on to the
> > > > name if the value is > 0. What do you think?
> > >
> > > Oh, you mean the "gadget.N" subdirs, which are the targets of the
> > > symlinks above? These were indeed named "gadget" before.
> > >
> > > Would it be possible to append the ".N" suffixes only to the actual
> > > symlinks, while keeping the target directory names unchanged?
> > > E.g. /sys/bus/gadget/devices/gadget.0 ->
> > > ../../../devices/platform/soc/e659c000.usb/gadget
> >
> > No, it's not possible. Or at least, not without significant changes to
> > the driver core. Besides, people expect these names to be the same.
>
> It should always be ok to change the names of devices as those are not
> going to be persistent / determinisitic. It's the attributes of devices
> that are supposed to be used to determine those types of things.
>
> So I think let's start out with the .N suffix for everything for now.
> I'll be glad to submit the fixed patch to the Android kernel build
> system to see what their testing comes back with to see if they happened
> to make any name assumptions as that's the largest user of this
> codebase.
Okay. Do you need me to send it as a proper patch before you try it
out?
Alan Stern
On Mon, May 09, 2022 at 12:47:19PM -0400, Alan Stern wrote:
> On Mon, May 09, 2022 at 06:23:31PM +0200, Greg KH wrote:
> > On Mon, May 09, 2022 at 11:05:06AM -0400, Alan Stern wrote:
> > > On Mon, May 09, 2022 at 04:42:01PM +0200, Geert Uytterhoeven wrote:
> > > > Hi Alan,
> > > >
> > > > On Mon, May 9, 2022 at 4:15 PM Alan Stern <[email protected]> wrote:
> > > > > On Mon, May 09, 2022 at 09:46:25AM +0200, Geert Uytterhoeven wrote:
> > > > > > > Geert:
> > > > > > >
> > > > > > > Can you test the patch below? It ought to fix the problem (although it
> > > > > >
> > > > > > Thanks!
> > > > > >
> > > > > > root@h3-salvator-xs:~# ls -l /sys/bus/gadget/devices/
> > > > > > total 0
> > > > > > lrwxrwxrwx 1 root root 0 Feb 14 2019 gadget.0 ->
> > > > > > ../../../devices/platform/soc/e659c000.usb/gadget.0
> > > > > > lrwxrwxrwx 1 root root 0 Feb 14 2019 gadget.1 ->
> > > > > > ../../../devices/platform/soc/ee020000.usb/gadget.1
> > > > > > lrwxrwxrwx 1 root root 0 Feb 14 2019 gadget.2 ->
> > > > > > ../../../devices/platform/soc/e6590000.usb/gadget.2
> > > > > >
> > > > > > Tested-by: Geert Uytterhoeven <[email protected]>
> > > > > >
> > > > > > LGTM, so
> > > > > > Reviewed-by: Geert Uytterhoeven <[email protected]>
> > > > >
> > > > > Thanks!
> > > > >
> > > > > > > might end up causing other problems down the line...)
> > > > > >
> > > > > > Can you please elaborate? I'm not too familiar with UBS gadgets.
> > > > >
> > > > > I was concerned about the fact that changing the name of a file,
> > > > > directory, or symbolic link in sysfs means changing a user API, and so
> > > > > it might cause some existing programs to fail. That would be a
> > > > > regression.
> > > > >
> > > > > Perhaps the best way to work around the problem is to leave the name set
> > > > > to "gadget" if the ID number is 0, while adding the ID number on to the
> > > > > name if the value is > 0. What do you think?
> > > >
> > > > Oh, you mean the "gadget.N" subdirs, which are the targets of the
> > > > symlinks above? These were indeed named "gadget" before.
> > > >
> > > > Would it be possible to append the ".N" suffixes only to the actual
> > > > symlinks, while keeping the target directory names unchanged?
> > > > E.g. /sys/bus/gadget/devices/gadget.0 ->
> > > > ../../../devices/platform/soc/e659c000.usb/gadget
> > >
> > > No, it's not possible. Or at least, not without significant changes to
> > > the driver core. Besides, people expect these names to be the same.
> >
> > It should always be ok to change the names of devices as those are not
> > going to be persistent / determinisitic. It's the attributes of devices
> > that are supposed to be used to determine those types of things.
> >
> > So I think let's start out with the .N suffix for everything for now.
> > I'll be glad to submit the fixed patch to the Android kernel build
> > system to see what their testing comes back with to see if they happened
> > to make any name assumptions as that's the largest user of this
> > codebase.
>
> Okay. Do you need me to send it as a proper patch before you try it
> out?
Yes please.
thanks,
greg k-h