Skip to contentSkip to conversation

console: usb: kbd: To improve TFTP booting performance

PATCH New Jim Lin 3 participants 6 messages Aug 13, 2013 – Aug 26, 2013

lore
Share & tools
lore

Jim Lin

author
#0 lore raw

PATCH v9 2/2 console: usb: kbd: To improve TFTP booting performance

TFTP booting is slow when a USB keyboard is installed and stdin has usbkbd added.
This fix is to change Ctrl-C polling for USB keyboard to every second when NET transfer is running.

Signed-off-by Jim Lin <[email protected]>

---
Changes in v2:
1. Change configuration name from CONFIG_CTRLC_POLL_MS to CONFIG_CTRLC_POLL_S.
2. New code will be executed only when CONFIG_CTRLC_POLL_S is defined in configuration header file.
3. Add description in README.console.
Changes in v3:
1. Move changes to common/usb_kbd.c and doc/README.usb
2. Rename config setting to CONFIG_USBKB_TESTC_PERIOD.
3. Remove slow response on USB-keyboard input when TFTP boot is not running.
Changes in v4:
1. Remove changes in doc/README.usb, common/usb_kbd.c and CONFIG_USBKB_TESTC_PERIOD
2. Modify net/net.c
Changes in v5:
1. Change variable name to ctrlc_t_start.
2. Use two calls of get_timer(0) to get time gap.
Changes in v6:
1. In common/usb_kbd.c, check net_busy_flag to determine whether we poll USB keyboard status.
2. In include/usb.h, add external variable declaration net_busy_flag
Changes in v7:
1. In common/usb_kbd.c and include/usb.h, add #ifdef CONFIG_CMD_NET.
2. In common/usb_kbd.c, modify code to get correct time gap.
Changes in v8:
1. Add __maybe_unused for variable kbd_testc_tms.
Changes in v9:
1. Move external variable declaration from include/usb.h to common/usb_kbd.c

 common/usb_kbd.c |   15 +++++++++++++++
 1 files changed, 15 insertions(+), 0 deletions(-)
common/usb_kbd.c +15 -0
index 3174b5e..46100e6 100644
@@ -121,6 +121,11 @@ struct usb_kbd_pdata {
 	uint8_t		flags;
 };
 
+extern int __maybe_unused net_busy_flag;
+
+/* The period of time between two calls of usb_kbd_testc(). */
+static unsigned long __maybe_unused kbd_testc_tms;
+
 /* Generic keyboard event polling. */
 void usb_kbd_generic_poll(void)
 {

@@ -366,6 +371,16 @@ static int usb_kbd_testc(void)
 	struct usb_device *usb_kbd_dev;
 	struct usb_kbd_pdata *data;
 
+#ifdef CONFIG_CMD_NET
+	/*
+	 * If net_busy_flag is 1, NET transfer is running,
+	 * then we check key-pressed every second (first check may be
+	 * less than 1 second) to improve TFTP booting performance.
+	 */
+	if (net_busy_flag && (get_timer(kbd_testc_tms) < CONFIG_SYS_HZ))
+		return 0;
+	kbd_testc_tms = get_timer(0);
+#endif
 	dev = stdio_get_by_name(DEVNAME);
 	usb_kbd_dev = (struct usb_device *)dev->priv;
 	data = usb_kbd_dev->privptr;
Signature · 1 line
1.7.7
Applied
#1 lore raw

PATCH v9 2/2 console: usb: kbd: To improve TFTP booting performance

Dear Jim Lin,

Quoted 36 lines

TFTP booting is slow when a USB keyboard is installed and stdin has usbkbd added.
This fix is to change Ctrl-C polling for USB keyboard to every second when NET transfer is running.

Signed-off-by Jim Lin <[email protected]>

---
Changes in v2:
1. Change configuration name from CONFIG_CTRLC_POLL_MS to
CONFIG_CTRLC_POLL_S. 2. New code will be executed only when CONFIG_CTRLC_POLL_S is defined in configuration header file.
3. Add description in README.console.
Changes in v3:
1. Move changes to common/usb_kbd.c and doc/README.usb
2. Rename config setting to CONFIG_USBKB_TESTC_PERIOD.
3. Remove slow response on USB-keyboard input when TFTP boot is not
running. Changes in v4:
1. Remove changes in doc/README.usb, common/usb_kbd.c and CONFIG_USBKB_TESTC_PERIOD
2. Modify net/net.c
Changes in v5:
1. Change variable name to ctrlc_t_start.
2. Use two calls of get_timer(0) to get time gap.
Changes in v6:
1. In common/usb_kbd.c, check net_busy_flag to determine whether we poll USB keyboard status.
2. In include/usb.h, add external variable declaration net_busy_flag
Changes in v7:
1. In common/usb_kbd.c and include/usb.h, add #ifdef CONFIG_CMD_NET.
2. In common/usb_kbd.c, modify code to get correct time gap.
Changes in v8:
1. Add __maybe_unused for variable kbd_testc_tms.
Changes in v9:
1. Move external variable declaration from include/usb.h to
common/usb_kbd.c

Quoted 2 lines
 common/usb_kbd.c |   15 +++++++++++++++
 1 files changed, 15 insertions(+), 0 deletions(-)

Applied, thanks.

Best regards,
Marek Vasut

author Applied
#2 lore raw

PATCH v9 2/2 console: usb: kbd: To improve TFTP booting performance

Marek,
In common/usb_kbd.c, you applied code to wrong place.
You should apply my change to function usb_kbd_testc(), instead of usb_kbd_getc(). Could you help to correct it?

Thanks,
Jim

-----Original Message-----
From: Marek Vasut [mailto:marex at denx.de]
Sent: Wednesday, August 21, 2013 12:27 PM
To: Jim Lin
Cc: joe.hershberger at gmail.com; u-boot at lists.denx.de; Tom Warren; swarren at wwwdotorg.org
Subject: Re: [PATCH v9 2/2] console: usb: kbd: To improve TFTP booting performance

Dear Jim Lin,

Quoted 36 lines

TFTP booting is slow when a USB keyboard is installed and stdin has usbkbd added.
This fix is to change Ctrl-C polling for USB keyboard to every second when NET transfer is running.

Signed-off-by Jim Lin <[email protected]>

---
Changes in v2:
1. Change configuration name from CONFIG_CTRLC_POLL_MS to
CONFIG_CTRLC_POLL_S. 2. New code will be executed only when CONFIG_CTRLC_POLL_S is defined in configuration header file.
3. Add description in README.console.
Changes in v3:
1. Move changes to common/usb_kbd.c and doc/README.usb 2. Rename
config setting to CONFIG_USBKB_TESTC_PERIOD.
3. Remove slow response on USB-keyboard input when TFTP boot is not
running. Changes in v4:
1. Remove changes in doc/README.usb, common/usb_kbd.c and CONFIG_USBKB_TESTC_PERIOD
2. Modify net/net.c
Changes in v5:
1. Change variable name to ctrlc_t_start.
2. Use two calls of get_timer(0) to get time gap.
Changes in v6:
1. In common/usb_kbd.c, check net_busy_flag to determine whether we poll USB keyboard status.
2. In include/usb.h, add external variable declaration net_busy_flag
Changes in v7:
1. In common/usb_kbd.c and include/usb.h, add #ifdef CONFIG_CMD_NET.
2. In common/usb_kbd.c, modify code to get correct time gap.
Changes in v8:
1. Add __maybe_unused for variable kbd_testc_tms.
Changes in v9:
1. Move external variable declaration from include/usb.h to
common/usb_kbd.c

Quoted 2 lines
 common/usb_kbd.c |   15 +++++++++++++++
 1 files changed, 15 insertions(+), 0 deletions(-)

Applied, thanks.

Best regards,
Marek Vasut

--
nvpublic

#3 lore raw

PATCH v9 2/2 console: usb: kbd: To improve TFTP booting performance

Dear Jim Lin,

Quoted 5 lines

Marek,
In common/usb_kbd.c, you applied code to wrong place. You should apply my change to function usb_kbd_testc(), instead of usb_kbd_getc(). Could you help to correct it?

I believe I applied your patches. Anyway, submit a subsequent patch against u- boot-usb/master please.

Best regards,
Marek Vasut

#4 lore raw

PATCH v9 2/2 console: usb: kbd: To improve TFTP booting performance

Quoted 10 lines · Marek Vasut

Dear Jim Lin,

Quoted 5 lines

Marek,
In common/usb_kbd.c, you applied code to wrong place. You should apply my change to function usb_kbd_testc(), instead of usb_kbd_getc(). Could you help to correct it?

I believe I applied your patches. Anyway, submit a subsequent patch against u- boot-usb/master please.

Marek, did you apply the patch using "git am", "patch", or by hand-editing? Looking at the fixup patch that Jim sent, the patch context is identical in both places, so perhaps this exposed some issue with git/... - it'd be useful to understand how the problem occurred, so it can be fixed.

#5 lore raw

PATCH v9 2/2 console: usb: kbd: To improve TFTP booting performance

Dear Stephen Warren,

Quoted 8 lines · Marek Vasut
Quoted 7 lines

Dear Jim Lin,

Quoted 4 lines

Marek,
In common/usb_kbd.c, you applied code to wrong place. You should apply my change to function usb_kbd_testc(), instead of usb_kbd_getc(). Could you help to correct it?

Quoted 8 lines
Quoted 2 lines

I believe I applied your patches. Anyway, submit a subsequent patch against u- boot-usb/master please.

Marek, did you apply the patch using "git am", "patch", or by hand-editing? Looking at the fixup patch that Jim sent, the patch context is identical in both places, so perhaps this exposed some issue with git/... - it'd be useful to understand how the problem occurred, so it can be fixed.

git am ... and you're right, this is somewhat strange.

Best regards,
Marek Vasut