[PATCH] Reset before reading Chip ID

[Sean Nelson]

Add a reset to probe_jedec before we read the Chip's IDs.
Signed-off-by: Sean Nelson

[Anders Juel Jensen]

On Sunday 31 January 2010 06:33:12 Sean Nelson wrote:
> Add a reset to probe_jedec before we read the Chip's IDs.
> Signed-off-by: Sean Nelson

Acked-by: Anders Juel Jensen

[Sean Nelson]

I'd like Uwe, Stepan, or Carl-Daniel to test this and send an ack/nack.

On 1/31/10 8:07 AM, Anders Juel Jensen wrote:
> On Sunday 31 January 2010 06:33:12 Sean Nelson wrote:
> > Add a reset to probe_jedec before we read the Chip's IDs.
> > Signed-off-by: Sean Nelson
>
> Acked-by: Anders Juel Jensen
>
> flashrom mailing list
> [e-mail]
> http://www.flashrom.org/mailman/listinfo/flashrom

[Carl-Daniel Hailfinger]

On 04.02.2010 07:13, Sean Nelson wrote:
> I'd like Uwe, Stepan, or Carl-Daniel to test this and send an ack/nack.
>
> On 1/31/10 8:07 AM, Anders Juel Jensen wrote:
> > On Sunday 31 January 2010 06:33:12 Sean Nelson wrote:
> > > Add a reset to probe_jedec before we read the Chip's IDs.
> > > Signed-off-by: Sean Nelson
> >
> > Acked-by: Anders Juel Jensen

Sorry for the long delay. We absolutely want this patch, and I'm happy
with it except for two minor issues:
- Previous probes might have had too short delays for entering ID mode,
so the chip may still be in the process of entering the ID mode. Due to
that, an additional delay before the reset makes sense.
- We needed FEATURE_RESET_MASK to deal cleanly with those feature bits.

Maciej Pijanka tested the patch and it fixes probing for him with some
old Atmel chips.

Signed-off-by: Sean Nelson 
Signed-off-by: Carl-Daniel Hailfinger 
Acked-by: Anders Juel Jensen

Index: flashrom-reset_before_probe/flash.h
===================================================================
--- flashrom-reset_before_probe/flash.h	(Revision 1172)
+++ flashrom-reset_before_probe/flash.h	(Arbeitskopie)
@@ -78,6 +78,7 @@
 #define FEATURE_LONG_RESET	(0 << 4)
 #define FEATURE_SHORT_RESET	(1 << 4)
 #define FEATURE_EITHER_RESET	FEATURE_LONG_RESET
+#define FEATURE_RESET_MASK	(FEATURE_LONG_RESET | FEATURE_SHORT_RESET)
 #define FEATURE_ADDR_FULL	(0 << 2)
 #define FEATURE_ADDR_MASK	(3 << 2)
 #define FEATURE_ADDR_2AA	(1 << 2)
Index: flashrom-reset_before_probe/jedec.c
===================================================================
--- flashrom-reset_before_probe/jedec.c	(Revision 1172)
+++ flashrom-reset_before_probe/jedec.c	(Arbeitskopie)
@@ -123,6 +123,26 @@
    	return 0;
    }

+	/* Earlier probes might have been too fast for the chip to enter ID
+	 * mode completely. Allow the chip to finish this before seeing a
+	 * reset command.
+	 */
+	if (probe_timing_enter)
+		programmer_delay(probe_timing_enter);
+	/* Reset chip to a clean slate */
+	if ((flash->feature_bits & FEATURE_RESET_MASK) == FEATURE_LONG_RESET)
+	{
+		chip_writeb(0xAA, bios + (0x5555 & mask));
+		if (probe_timing_exit)
+			programmer_delay(10);
+		chip_writeb(0x55, bios + (0x2AAA & mask));
+		if (probe_timing_exit)
+			programmer_delay(10);
+	}
+	chip_writeb(0xF0, bios + (0x5555 & mask));
+	if (probe_timing_exit)
+		programmer_delay(probe_timing_exit);
+
    /* Issue JEDEC Product ID Entry command */
    chip_writeb(0xAA, bios + (0x5555 & mask));
    if (probe_timing_enter)
@@ -153,7 +173,7 @@
    }

/* Issue JEDEC Product ID Exit command */
-	if ((flash->feature_bits & FEATURE_SHORT_RESET) == FEATURE_LONG_RESET)
+	if ((flash->feature_bits & FEATURE_RESET_MASK) == FEATURE_LONG_RESET)
    {
    	chip_writeb(0xAA, bios + (0x5555 & mask));
    	if (probe_timing_exit)

[Maciej Pijanka]

On Wed, 15 Sep 2010, Carl-Daniel Hailfinger wrote:
> On 04.02.2010 07:13, Sean Nelson wrote:
> > I'd like Uwe, Stepan, or Carl-Daniel to test this and send an ack/nack.
> >
> > On 1/31/10 8:07 AM, Anders Juel Jensen wrote:
> > > On Sunday 31 January 2010 06:33:12 Sean Nelson wrote:
> > > > Add a reset to probe_jedec before we read the Chip's IDs.
> > > > Signed-off-by: Sean Nelson
> > >
> > > Acked-by: Anders Juel Jensen
>
> Sorry for the long delay. We absolutely want this patch, and I'm happy
> with it except for two minor issues:
> Previous probes might have had too short delays for entering ID mode,
>
> so the chip may still be in the process of entering the ID mode. Due to
> that, an additional delay before the reset makes sense.
> We needed FEATURE_RESET_MASK to deal cleanly with those feature bits.
>
> Maciej Pijanka tested the patch and it fixes probing for him with some
> old Atmel chips.

I checked current version of patch too, works here with nic3com and
AT29C512 in socket.

[Carl-Daniel Hailfinger]

On 15.09.2010 18:19, Maciej Pijanka wrote:
> On Wed, 15 Sep 2010, Carl-Daniel Hailfinger wrote:
> > Previous probes might have had too short delays for entering ID mode,
> >
> > so the chip may still be in the process of entering the ID mode. Due to
> > that, an additional delay before the reset makes sense.
> > We needed FEATURE_RESET_MASK to deal cleanly with those feature bits.
> >
> > Maciej Pijanka tested the patch and it fixes probing for him with some
> > old Atmel chips.
>
> I checked current version of patch too, works here with nic3com and
> AT29C512 in socket.

Thanks, committed in r1216.

Regards,
Carl-Daniel

[Sean Nelson]

On 9/15/10 4:23 AM, Carl-Daniel Hailfinger wrote:
> On 04.02.2010 07:13, Sean Nelson wrote:
> > I'd like Uwe, Stepan, or Carl-Daniel to test this and send an ack/nack.
> >
> > On 1/31/10 8:07 AM, Anders Juel Jensen wrote:
> > > On Sunday 31 January 2010 06:33:12 Sean Nelson wrote:
> > > > Add a reset to probe_jedec before we read the Chip's IDs.
> > > > Signed-off-by: Sean Nelson
> > >
> > > Acked-by: Anders Juel Jensen
>
> Sorry for the long delay. We absolutely want this patch, and I'm happy
> with it except for two minor issues:
> Previous probes might have had too short delays for entering ID mode,
>
> so the chip may still be in the process of entering the ID mode. Due to
> that, an additional delay before the reset makes sense.
> We needed FEATURE_RESET_MASK to deal cleanly with those feature bits.
>
> Maciej Pijanka tested the patch and it fixes probing for him with some
> old Atmel chips.
>
> Signed-off-by: Sean Nelson
> Signed-off-by: Carl-Daniel Hailfinger
> Acked-by: Anders Juel Jensen

Acked-by: Sean Nelson