Skip to contentSkip to conversation

ntp: make is_error_status() use its argument

PATCH In review George Spelvin 3 participants 5 messages May 12, 2014 – May 13, 2014 Timekeeping, clocksource core, NTP, alarmtimer

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 ntp: make is_error_status() use its argument

It's an inline function always called with the global time_status as an argument, so there's zero functional difference, but the non-CONFIG_SMP version uses the passed-in argument, while the CONFIG_SMP one ignores its argument and uses the global.

Make it use the argument always; shorter variable names are good.

---
While poking about in the code, I came across this rather odd bit. It looked worth fixing, on general principles.

kernel/time/ntp.c +6 -6
index 419a52cecd..1a2aad3fff 100644
@@ -165,21 +165,21 @@ static inline void pps_set_freq(s64 freq)
 
 static inline int is_error_status(int status)
 {
-	return (time_status & (STA_UNSYNC|STA_CLOCKERR))
+	return (status & (STA_UNSYNC|STA_CLOCKERR))
 		/* PPS signal lost when either PPS time or
 		 * PPS frequency synchronization requested
 		 */
-		|| ((time_status & (STA_PPSFREQ|STA_PPSTIME))
-			&& !(time_status & STA_PPSSIGNAL))
+		|| ((status & (STA_PPSFREQ|STA_PPSTIME))
+			&& !(status & STA_PPSSIGNAL))
 		/* PPS jitter exceeded when
 		 * PPS time synchronization requested */
-		|| ((time_status & (STA_PPSTIME|STA_PPSJITTER))
+		|| ((status & (STA_PPSTIME|STA_PPSJITTER))
 			== (STA_PPSTIME|STA_PPSJITTER))
 		/* PPS wander exceeded or calibration error when
 		 * PPS frequency synchronization requested
 		 */
-		|| ((time_status & STA_PPSFREQ)
-			&& (time_status & (STA_PPSWANDER|STA_PPSERROR)));
+		|| ((status & STA_PPSFREQ)
+			&& (status & (STA_PPSWANDER|STA_PPSERROR)));
 }
 
 static inline void pps_fill_timex(struct timex *txc)
#1 lore raw
Quoted 4 lines · George Spelvin

It's an inline function always called with the global time_status as an argument, so there's zero functional difference, but the non-CONFIG_SMP version uses the passed-in argument, while the CONFIG_SMP one ignores its argument and uses the global.

CONFIG_NTP_PPS not CONFIG_SMP, right?

Good catch though, looks like the check code was re-factored out, but someone forgot to use the local variable.

Adding Alexander since he submitted the pps logic.

If there's no objections, I'll queue this (with a more verbose commit message) for 3.16

thanks
-john

Quoted 40 lines

Make it use the argument always; shorter variable names are good.

---
While poking about in the code, I came across this rather odd bit. It looked worth fixing, on general principles.

kernel/time/ntp.c +6 -6
index 419a52cecd..1a2aad3fff 100644
@@ -165,21 +165,21 @@ static inline void pps_set_freq(s64 freq)
 
 static inline int is_error_status(int status)
 {
-	return (time_status & (STA_UNSYNC|STA_CLOCKERR))
+	return (status & (STA_UNSYNC|STA_CLOCKERR))
 		/* PPS signal lost when either PPS time or
 		 * PPS frequency synchronization requested
 		 */
-		|| ((time_status & (STA_PPSFREQ|STA_PPSTIME))
-			&& !(time_status & STA_PPSSIGNAL))
+		|| ((status & (STA_PPSFREQ|STA_PPSTIME))
+			&& !(status & STA_PPSSIGNAL))
 		/* PPS jitter exceeded when
 		 * PPS time synchronization requested */
-		|| ((time_status & (STA_PPSTIME|STA_PPSJITTER))
+		|| ((status & (STA_PPSTIME|STA_PPSJITTER))
 			== (STA_PPSTIME|STA_PPSJITTER))
 		/* PPS wander exceeded or calibration error when
 		 * PPS frequency synchronization requested
 		 */
-		|| ((time_status & STA_PPSFREQ)
-			&& (time_status & (STA_PPSWANDER|STA_PPSERROR)));
+		|| ((status & STA_PPSFREQ)
+			&& (status & (STA_PPSWANDER|STA_PPSERROR)));
 }
 
 static inline void pps_fill_timex(struct timex *txc)
#2 lore raw

No objections from me, of course. Good catch.

В Mon, 12 May 2014 10:54:29 -0700
John Stultz <[email protected]> пишет:

Quoted 62 lines · George Spelvin
Quoted 4 lines

It's an inline function always called with the global time_status as an argument, so there's zero functional difference, but the non-CONFIG_SMP version uses the passed-in argument, while the CONFIG_SMP one ignores its argument and uses the global.

CONFIG_NTP_PPS not CONFIG_SMP, right?

Good catch though, looks like the check code was re-factored out, but someone forgot to use the local variable.

Adding Alexander since he submitted the pps logic.

If there's no objections, I'll queue this (with a more verbose commit message) for 3.16

thanks
-john

Quoted 41 lines

Make it use the argument always; shorter variable names are good.

---
While poking about in the code, I came across this rather odd bit. It looked worth fixing, on general principles.

kernel/time/ntp.c +6 -6
index 419a52cecd..1a2aad3fff 100644
@@ -165,21 +165,21 @@ static inline void pps_set_freq(s64 freq)
 
 static inline int is_error_status(int status)
 {
-	return (time_status & (STA_UNSYNC|STA_CLOCKERR))
+	return (status & (STA_UNSYNC|STA_CLOCKERR))
 		/* PPS signal lost when either PPS time or
 		 * PPS frequency synchronization requested
 		 */
-		|| ((time_status & (STA_PPSFREQ|STA_PPSTIME))
-			&& !(time_status & STA_PPSSIGNAL))
+		|| ((status & (STA_PPSFREQ|STA_PPSTIME))
+			&& !(status & STA_PPSSIGNAL))
 		/* PPS jitter exceeded when
 		 * PPS time synchronization requested */
-		|| ((time_status & (STA_PPSTIME|STA_PPSJITTER))
+		|| ((status & (STA_PPSTIME|STA_PPSJITTER))
 			== (STA_PPSTIME|STA_PPSJITTER))
 		/* PPS wander exceeded or calibration error when
 		 * PPS frequency synchronization requested
 		 */
-		|| ((time_status & STA_PPSFREQ)
-			&& (time_status &
 (STA_PPSWANDER|STA_PPSERROR)));
+		|| ((status & STA_PPSFREQ)
+			&& (status &
 (STA_PPSWANDER|STA_PPSERROR))); }
 

static inline void pps_fill_timex(struct timex *txc)

Signature · 1 line
  Alexander
#3 lore raw
In reply to John Stultz
Quoted 1 line

CONFIG_NTP_PPS not CONFIG_SMP, right?

D'oh, yes. I just noticed that there were two versions and one was broken. Somehow my search for the enclosing #ifdef was completely broken. (I could swear I saw CONFIG_SMP, but looking now it's not even mentioned in the file; maybe I switched editor windows?)

Quoted 1 line

Adding Alexander since he submitted the pps logic.

Thank you; some of these issues of code provenance and "who the heck do I notify about this?" are a bit laborious to infer even with the help of the git logs.

#4 lore raw
Quoted 5 lines · George Spelvin
Quoted 1 line

CONFIG_NTP_PPS not CONFIG_SMP, right?

D'oh, yes. I just noticed that there were two versions and one was broken. Somehow my search for the enclosing #ifdef was completely broken. (I could swear I saw CONFIG_SMP, but looking now it's not even mentioned in the file; maybe I switched editor windows?)

No worries. :)

Quoted 5 lines
Quoted 2 lines

Adding Alexander since he submitted the pps logic.

Thank you; some of these issues of code provenance and "who the heck do I notify about this?" are a bit laborious to infer even with the help of the git logs.

tig blame <filename> is pretty amazing for this sort of thing.

thanks
-john