Change in flashrom[master]: [RFC] udelay: Relax clock_gettime() resolution expectations

[Nico Huber (Code Review)]

Nico Huber has uploaded this change for review. ( https://review.coreboot.org/c/flashrom/+/39864 )

Change subject: [RFC] udelay: Relax clock_gettime() resolution expectations
......................................................................

[RFC] udelay: Relax clock_gettime() resolution expectations

Only few programmers actually want to delay by a single microsecond. So
we can relax the expected mininum resolution by default and let indivi-
dual programmers override as needed.

UNTESTED

Change-Id: Ia24f5c8440b7cb8abcc5a1509443d0c41fd2c559
Signed-off-by: Nico Huber 
---
M bitbang_spi.c
M cli_classic.c
M nicintel_eeprom.c
M udelay.c
4 files changed, 15 insertions(+), 7 deletions(-)

git pull ssh://review.coreboot.org:29418/flashrom refs/changes/64/39864/1

diff --git a/bitbang_spi.c b/bitbang_spi.c
index 7c183c3..8e5b8f2 100644
--- a/bitbang_spi.c
+++ b/bitbang_spi.c
@@ -104,6 +104,8 @@
    mst.data = master;
    register_spi_master(&mst);

+	myusec_calibrate_delay(100);
+
    /* Only mess with the bus if we're sure nobody else uses it. */
    bitbang_spi_request_bus(master);
    bitbang_spi_set_cs(master, 1);
diff --git a/cli_classic.c b/cli_classic.c
index 73cc417..2436351 100644
--- a/cli_classic.c
+++ b/cli_classic.c
@@ -454,7 +454,7 @@
    }

/* FIXME: Delay calibration should happen in programmer code. */
-	myusec_calibrate_delay();
+	myusec_calibrate_delay(1000);

if (programmer_init(prog, pparam)) {
    	msg_perr("Error: Programmer initialization failed.\n");
diff --git a/nicintel_eeprom.c b/nicintel_eeprom.c
index 4d45f79..06284d4 100644
--- a/nicintel_eeprom.c
+++ b/nicintel_eeprom.c
@@ -449,6 +449,8 @@

int nicintel_ee_init(void)
 {
+	myusec_calibrate_delay(100);
+
    if (rget_io_perms())
    	return 1;

diff --git a/udelay.c b/udelay.c
index 6c0efc4..3f0b6d1 100644
--- a/udelay.c
+++ b/udelay.c
@@ -52,11 +52,11 @@
    } while (now.tv_sec < end.tv_sec || (now.tv_sec == end.tv_sec && now.tv_nsec < end.tv_nsec));
 }

-static int clock_check_res(void)
+static int clock_check_res(const unsigned int resolution_ns)
 {
    struct timespec res;
    if (!clock_getres(clock_id, &res)) {
-		if (res.tv_sec == 0 && res.tv_nsec <= 100) {
+		if (res.tv_sec == 0 && res.tv_nsec <= resolution_ns) {
    		msg_pinfo("Using clock_gettime for delay loops (clk_id: %d, resolution: %ldns).\n",
    			  (int)clock_id, res.tv_nsec);
    		use_clock_gettime = true;
@@ -65,7 +65,7 @@
    } else if (clock_id != CLOCK_REALTIME && errno == EINVAL) {
    	/* Try again with CLOCK_REALTIME. */
    	clock_id = CLOCK_REALTIME;
-		return clock_check_res();
+		return clock_check_res(resolution_ns);
    }
    return 0;
 }
@@ -77,7 +77,7 @@
 #endif /* HAVE_CLOCK_GETTIME == 1 */

/* loops per microsecond */
-static unsigned long micro = 1;
+static unsigned long micro;

__attribute__ ((noinline)) void myusec_delay(unsigned int usecs)
 {
@@ -133,15 +133,19 @@
    return timeusec;
 }

-void myusec_calibrate_delay(void)
+void myusec_calibrate_delay(const unsigned int resolution_ns)
 {
-	if (clock_check_res())
+	if (clock_check_res(resolution_ns))
    	return;

+	if (micro)
+		return;	/* already calibrated */
+
    unsigned long count = 1000;
    unsigned long timeusec, resolution;
    int i, tries = 0;

+	micro = 1;
    msg_pinfo("Calibrating delay loop... ");
    resolution = measure_os_delay_resolution();
    if (resolution) {

[Nico Huber (Code Review)]

Hello build bot (Jenkins),

I'd like you to reexamine a change. Please visit

https://review.coreboot.org/c/flashrom/+/39864

to look at the new patch set (#2).

Change subject: [RFC] udelay: Relax clock_gettime() resolution expectations
......................................................................

[RFC] udelay: Relax clock_gettime() resolution expectations

Only few programmers actually want to delay by a single microsecond. So
we can relax the expected mininum resolution by default and let indivi-
dual programmers override as needed.

UNTESTED

Change-Id: Ia24f5c8440b7cb8abcc5a1509443d0c41fd2c559
Signed-off-by: Nico Huber 
---
M bitbang_spi.c
M cli_classic.c
M libflashrom.c
M nicintel_eeprom.c
M programmer.h
M udelay.c
6 files changed, 17 insertions(+), 9 deletions(-)

git pull ssh://review.coreboot.org:29418/flashrom refs/changes/64/39864/2

[Nico Huber (Code Review)]

Hello build bot (Jenkins),

I'd like you to reexamine a change. Please visit

https://review.coreboot.org/c/flashrom/+/39864

to look at the new patch set (#3).

Change subject: [RFC] udelay: Relax clock_gettime() resolution expectations
......................................................................

[RFC] udelay: Relax clock_gettime() resolution expectations

Only few programmers actually want to delay by a single microsecond. So
we can relax the expected mininum resolution by default and let indivi-
dual programmers override as needed.

UNTESTED

Change-Id: Ia24f5c8440b7cb8abcc5a1509443d0c41fd2c559
Signed-off-by: Nico Huber 
---
M bitbang_spi.c
M cli_classic.c
M libflashrom.c
M nicintel_eeprom.c
M programmer.h
M udelay.c
6 files changed, 17 insertions(+), 9 deletions(-)

git pull ssh://review.coreboot.org:29418/flashrom refs/changes/64/39864/3