http://blog.gmane.org/gmane.linux.rt.user hourly 1 1901-01-01T00:00+00:00 http://gmane.org/img/gmane-25t.png http://gmane.org http://comments.gmane.org/gmane.linux.rt.user/8354 <pre>This is a preliminary hack, I need to clean it up a little, but incase you want to give it a try, here is the alpha version. Note - I stole this methodology from tools/perf Signed-off-by: John Kacur &lt;jkacur&lt; at &gt;redhat.com&gt; --- Makefile | 33 +++++++++++++++++++++++++++------ config/feature-tests.mak | 8 ++++++++ config/utilities.mak | 7 +++++++ 3 files changed, 42 insertions(+), 6 deletions(-) create mode 100644 config/feature-tests.mak create mode 100644 config/utilities.mak diff --git a/Makefile b/Makefile index 3a82407..33f5027 100644 --- a/Makefile +++ b/Makefile &lt; at &gt;&lt; at &gt; -16,9 +16,6 &lt; at &gt;&lt; at &gt; srcdir?= $(prefix)/src machinetype = $(shell $(CC) -dumpmachine | \ sed -e 's/-.*//' -e 's/i.86/i386/' -e 's/mips.*/mips/' -e 's/ppc.*/powerpc/') -ifneq ($(filter x86_64 i386 ia64 mips powerpc,$(machinetype)),) -NUMA := 1 -endif CFLAGS ?= -D_GNU_SOURCE -Wall -Wno-nonnull -Isrc/include LDFLAGS ?= &lt; at &gt;&lt; at &gt; -31,9 +28,33 &lt; at &gt;&lt; at &gt; else CFLAGS+= -O0 -g endif -ifeq ($(NUMA),1) -CFLAGS += -DNUMA -NUMA_LIBS = -lnuma +ifeq ("$(origin O)", "command line") +OUTPUT := $(O)/ +endif + +ifneq ($(OUTPUT),) +# check that the output directory actually exists +OUTDIR := $(shell cd $(OUTPUT) &amp;&amp; /bin/pwd) +$(if $(OUTDIR),, $(error output directory "$(OUTPUT)" does not exist)) +endif + +include config/utilities.mak +include config/feature-tests.mak + +ifeq ($(NUMA),0) +NO_NUMA=0 +endif + +ifdef NO_NUMA +CFLAGS += -DNO_NUMA_SUPPORT +else +ifneq ($(call try-cc,$(SOURCE_NUMA),-lnuma),y) +msg := $(warning numa not found. Install numactl-devel to build with numa support); +CFLAGS += -DNO_NUMA_SUPPORT +else +CFLAGS += -DNUMA +NUMA_LIBS = -lnuma +endif endif VPATH= src/cyclictest: diff --git a/config/feature-tests.mak b/config/feature-tests.mak new file mode 100644 index 0000000..8fb61ea --- /dev/null +++ b/config/feature-tests.mak &lt; at &gt;&lt; at &gt; -0,0 +1,8 &lt; at &gt;&lt; at &gt; +ifndef NO_NUMA +define SOURCE_NUMA +int main(void) +{ +return numa_available(); +} +endef +endif diff --git a/config/utilities.mak b/config/utilities.mak new file mode 100644 index 0000000..da7c434 --- /dev/null +++ b/config/utilities.mak &lt; at &gt;&lt; at &gt; -0,0 +1,7 &lt; at &gt;&lt; at &gt; +# try-cc +# Usage: option = $(call try-cc, source-to-build, cc-options) +try-cc = $(shell sh -c \ +'TMP="$(OUTPUT)$(TMPOUT).$$$$"; \ + echo "$(1)" | \ + $(CC) -x c - $(2) -o "$$TMP" &gt; /dev/null 2&gt;&amp;1 &amp;&amp; echo y; \ + rm -f "$$TMP"') </pre> John Kacur 2012-05-25T23:39:49 http://comments.gmane.org/gmane.linux.rt.user/8348 <pre> Dear RT Folks, I'm pleased to announce the 3.0.32-rt52 stable release. This release is just an update to the new stable 3.0.32 version and no RT specific changes have been made. You can get this release via the git tree at: git://git.kernel.org/pub/scm/linux/kernel/git/rt/linux-stable-rt.git Head SHA1: 63166b1a384148473180f7bef83e71409e38de14 Or to build 3.0.32-rt52 directly, the following patches should be applied: http://www.kernel.org/pub/linux/kernel/v3.0/linux-3.0.tar.xz http://www.kernel.org/pub/linux/kernel/v3.0/patch-3.0.32.xz http://www.kernel.org/pub/linux/kernel/projects/rt/3.0/patch-3.0.32-rt52.patch.xz Enjoy, </pre> Steven Rostedt 2012-05-22T20:19:36 http://comments.gmane.org/gmane.linux.rt.user/8341 <pre> Dear RT Folks, I'm pleased to announce the 3.2.18-rt29 stable release. This release is just an update to the new stable 3.2.18 version and no RT specific changes have been made. You can get this release via the git tree at: git://git.kernel.org/pub/scm/linux/kernel/git/rt/linux-stable-rt.git Head SHA1: c96da77dcbdad05e0d707f443328e708f7cb024a Or to build 3.2.18-rt29 directly, the following patches should be applied: http://www.kernel.org/pub/linux/kernel/v3.x/linux-3.2.tar.xz http://www.kernel.org/pub/linux/kernel/v3.x/patch-3.2.18.xz http://www.kernel.org/pub/linux/kernel/projects/rt/3.2/patch-3.2.18-rt29.patch.xz Enjoy, </pre> Steven Rostedt 2012-05-22T15:01:48 http://comments.gmane.org/gmane.linux.rt.user/8326 <pre> Hello everybody out there! I use OSADL last stable real-time kernel from Pengutronix (&lt;http://pengutronix.de/software/linux-rt/debian_en.html&gt;) on a computer which configuration is given in the end of this message. On this computer, I am using Debian Squeeze (6.0), and, because I have to do some live recording combined with MIDI playing, I need to use a real-time kernel. As I am using Debian stable, I want to use the most stable kernel. Therefore, I have chosen OSADL kernel. But I am experiencing some troubles, linked with my graphic card. Tested on the classic kernel, I need to install “firmware-linux-nonfree” and the proprietary Catalyst driver for my graphic card. If not, then power consumption is too important, then too much heat is dissipated and display freezes after few minutes. To install the proprietary driver, I have done the following: # aptitude install fglrx-source fglrx-driver fglrx-glx fglrx-control fglrx-atieventsd module-assistant build-essential linux-headers-2.6 # module-assistant a-i fglrx # aticonfig --inital When I use the classic kernel, everything goes fine. During the installation of the real-time kernel, the configuration of the package which should replace “firmware-linux-nonfree” – unfortunately, I have forgotten its name – have failed. Also, when I run the real-time kernel, it is the open-source driver that is in use. As a consequence, after few minutes, the display freezes. I guess the first step is to configure the equivalent of “firmware-linux-nonfree”. Unfortunately, I do not have a clue on how to do this. As Pengutronix advises to use this mailing list, I come here to ask if anybody can help me in achieving this. Regards. Yoann The configuration of my computer: – Case: Fractal 3R; – power supply: Corsair VX 450W; – fan: 120mm Noctua NF-S12B ULN; – motherboard: Gigabyte H55M-USB3 (µATX LGA1156+DDR3+PCI-E); – processor: Intel Core i5 760 (2.8 QuadCore LGA1156); – radiator: Noctua NH-U9B SE2 775/1156/1366; – memory: 2x2 = 4 Go DDR3-1333 PC10600 Kingston; – graphic card: MSI R4350-MD512H/D3 Radeon HD 4350 512MB; – audio card: M-AUDIO Delta Audiophile 24/96 PCI; – SSD unit: 120 Go SATA SSD Intel X25-M Postville MLC; – hard-drive: 1.5 To SATA WD Green 64 Mo cache; – CD/DVD/Blu-Ray drive: burner CD/DVD LG Black 10X BD-ROM 16X DVD-ROM. </pre> Yoann LE BARS 2012-05-18T22:54:02 http://comments.gmane.org/gmane.linux.rt.user/8320 <pre>Not sure whether Real time micro benchmark suite is still being maintained at http://sourceforge.net/projects/rtmicrobench. The latest code from sourceforge has issues while running rtmb_c tests. It fails complaining about not being able to set thread priority. Here's a fix for it. https://github.com/vsubbiah/rtmb/commit/2a6714030a4009b50b76e99a83bf4b042cc82780 Thanks, Venkat -- To unsubscribe from this list: send the line "unsubscribe linux-rt-users" in the body of a message to majordomo&lt; at &gt;vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html </pre> Venkat Subbiah 2012-05-17T21:15:52 http://comments.gmane.org/gmane.linux.rt.user/8310 <pre>From: Yong Zhang &lt;yong.zhang&lt; at &gt;windriver.com&gt; Because __cpu_disable is called in atomic context and spinlock is a mutex on -rt. Signed-off-by: Yong Zhang &lt;yong.zhang0&lt; at &gt;gmail.com&gt; --- arch/mips/cavium-octeon/smp.c | 6 +++--- 1 files changed, 3 insertions(+), 3 deletions(-) diff --git a/arch/mips/cavium-octeon/smp.c b/arch/mips/cavium-octeon/smp.c index ef9c34a..473c72b 100644 --- a/arch/mips/cavium-octeon/smp.c +++ b/arch/mips/cavium-octeon/smp.c &lt; at &gt;&lt; at &gt; -257,7 +257,7 &lt; at &gt;&lt; at &gt; DEFINE_PER_CPU(int, cpu_state); extern void fixup_irqs(void); -static DEFINE_SPINLOCK(smp_reserve_lock); +static DEFINE_RAW_SPINLOCK(smp_reserve_lock); static int octeon_cpu_disable(void) { &lt; at &gt;&lt; at &gt; -266,7 +266,7 &lt; at &gt;&lt; at &gt; static int octeon_cpu_disable(void) if (cpu == 0) return -EBUSY; -spin_lock(&amp;smp_reserve_lock); +raw_spin_lock(&amp;smp_reserve_lock); set_cpu_online(cpu, false); cpu_clear(cpu, cpu_callin_map); &lt; at &gt;&lt; at &gt; -277,7 +277,7 &lt; at &gt;&lt; at &gt; static int octeon_cpu_disable(void) flush_cache_all(); local_flush_tlb_all(); -spin_unlock(&amp;smp_reserve_lock); +raw_spin_unlock(&amp;smp_reserve_lock); return 0; } </pre> Yong Zhang 2012-05-17T10:15:09 http://comments.gmane.org/gmane.linux.rt.user/8309 <pre>1)enqueue_to_backlog() (called from netif_rx) should be bind to a particluar CPU. This can be achieved by disabling migration. No need to disable preemption 2)Fixes crash "BUG: scheduling while atomic: ksoftirqd" in case of RT. If preemption is disabled, enqueue_to_backog() is called in atomic context. And if backlog exceeds its count, kfree_skb() is called. But in RT, kfree_skb() might gets scheduled out, so it expects non atomic context. 3)When CONFIG_PREEMPT_RT_FULL is not defined, migrate_enable(), migrate_disable() maps to preempt_enable() and preempt_disable(), so no change in functionality in case of non-RT. -Replace preempt_enable(), preempt_disable() with migrate_enable(), migrate_disable() respectively -Replace get_cpu(), put_cpu() with get_cpu_light(), put_cpu_light() respectively Signed-off-by: Priyanka Jain &lt;Priyanka.Jain&lt; at &gt;freescale.com&gt; Acked-by: Rajan Srivastava &lt;Rajan.Srivastava&lt; at &gt;freescale.com&gt; --- Testing: Tested successfully on p4080ds(8-core SMP system) net/core/dev.c | 8 ++++---- 1 files changed, 4 insertions(+), 4 deletions(-) diff --git a/net/core/dev.c b/net/core/dev.c index 452db70..4017820 100644 --- a/net/core/dev.c +++ b/net/core/dev.c &lt; at &gt;&lt; at &gt; -2940,7 +2940,7 &lt; at &gt;&lt; at &gt; int netif_rx(struct sk_buff *skb) struct rps_dev_flow voidflow, *rflow = &amp;voidflow; int cpu; -preempt_disable(); +migrate_disable(); rcu_read_lock(); cpu = get_rps_cpu(skb-&gt;dev, skb, &amp;rflow); &lt; at &gt;&lt; at &gt; -2950,13 +2950,13 &lt; at &gt;&lt; at &gt; int netif_rx(struct sk_buff *skb) ret = enqueue_to_backlog(skb, cpu, &amp;rflow-&gt;last_qtail); rcu_read_unlock(); -preempt_enable(); +migrate_enable(); } else #endif { unsigned int qtail; -ret = enqueue_to_backlog(skb, get_cpu(), &amp;qtail); -put_cpu(); +ret = enqueue_to_backlog(skb, get_cpu_light(), &amp;qtail); +put_cpu_light(); } return ret; } </pre> Priyanka Jain 2012-05-17T04:05:11 http://comments.gmane.org/gmane.linux.rt.user/8297 <pre> V3: unchanged from V2 cyclictest getopt_long() parameter clean up. Clean up before following patch which will add a new option. Some elements of long_options were not in alphabetical order. Some elements of optstring were not in alphabetical order. '-e', '--latency' was missing help text short form of --duration ('D') was missing from optstring Change a few instances of leading spaces to tabs. Add white space to long_options to improve readability. Some cases of the switch processing the result of getopt_long() were not in alphabetical order. Did _not_ clean up option value parsing and processing. Signed-off-by: Frank Rowand &lt;frank.rowand&lt; at &gt;am.sony.com&gt; --- src/cyclictest/cyclictest.c | 143 73 +70 -0 ! 1 file changed, 73 insertions(+), 70 deletions(-) Index: b/src/cyclictest/cyclictest.c =================================================================== --- a/src/cyclictest/cyclictest.c +++ b/src/cyclictest/cyclictest.c &lt; at &gt;&lt; at &gt; -923,10 +923,11 &lt; at &gt;&lt; at &gt; static void display_help(int error) "-D --duration=t specify a length for the test run\n" " default is in seconds, but 'm', 'h', or 'd' maybe added\n" " to modify value to minutes, hours or days\n" + "-e --latency=PM_QOS write PM_QOS to /dev/cpu_dma_latency\n" "-E --event event tracing (used with -b)\n" "-f --ftrace function trace (when -b is active)\n" "-h --histogram=US dump a latency histogram to stdout after the run\n" - " (with same priority about many threads)\n" + " (with same priority about many threads)\n" " US is the max time to be be tracked in microseconds\n" "-H --histofall=US same as -h except with an additional summary column\n" "-i INTV --interval=INTV base interval of thread in us default=1000\n" &lt; at &gt;&lt; at &gt; -944,6 +945,8 &lt; at &gt;&lt; at &gt; static void display_help(int error) "-Q --priospread spread priority levels starting at specified value\n" "-r --relative use relative timer instead of absolute\n" "-s --system use sys_nanosleep and sys_setitimer\n" + "-S --smp Standard SMP testing: options -a -t -n and\n" + " same priority of all threads\n" "-t --threads one thread per available processor\n" "-t [NUM] --threads=NUM number of threads:\n" " without NUM, threads = max_cpus\n" &lt; at &gt;&lt; at &gt; -951,16 +954,14 &lt; at &gt;&lt; at &gt; static void display_help(int error) "-T TRACE --tracer=TRACER set tracing function\n" " configured tracers: %s\n" "-u --unbuffered force unbuffered output for live processing\n" + "-U --numa Standard NUMA testing (similar to SMP option)\n" + " thread data structures allocated from local node\n" "-v --verbose output values on stdout for statistics\n" " format: n:c:v n=tasknum c=count v=value in us\n" - "-w --wakeup task wakeup tracing (used with -b)\n" - "-W --wakeuprt rt task wakeup tracing (used with -b)\n" - "-y POLI --policy=POLI policy of realtime thread, POLI may be fifo(default) or rr\n" - " format: --policy=fifo(default) or --policy=rr\n" - "-S --smp Standard SMP testing: options -a -t -n and\n" - " same priority of all threads\n" - "-U --numa Standard NUMA testing (similar to SMP option)\n" - " thread data structures allocated from local node\n", + "-w --wakeup task wakeup tracing (used with -b)\n" + "-W --wakeuprt rt task wakeup tracing (used with -b)\n" + "-y POLI --policy=POLI policy of realtime thread, POLI may be fifo(default) or rr\n" + " format: --policy=fifo(default) or --policy=rr\n", tracers ); if (error) &lt; at &gt;&lt; at &gt; -1044,48 +1045,51 &lt; at &gt;&lt; at &gt; static void process_options (int argc, c for (;;) { int option_index = 0; -/** Options for getopt */ +/* + * Options for getopt + * Ordered alphabetically by single letter name + */ static struct option long_options[] = { -{"affinity", optional_argument, NULL, 'a'}, -{"breaktrace", required_argument, NULL, 'b'}, -{"preemptirqs", no_argument, NULL, 'B'}, -{"clock", required_argument, NULL, 'c'}, -{"context", no_argument, NULL, 'C'}, -{"distance", required_argument, NULL, 'd'}, -{"event", no_argument, NULL, 'E'}, -{"ftrace", no_argument, NULL, 'f'}, -{"histogram", required_argument, NULL, 'h'}, -{"histofall", required_argument, NULL, 'H'}, -{"interval", required_argument, NULL, 'i'}, -{"irqsoff", no_argument, NULL, 'I'}, -{"loops", required_argument, NULL, 'l'}, -{"mlockall", no_argument, NULL, 'm' }, -{"refresh_on_max", no_argument, NULL, 'M' }, -{"nanosleep", no_argument, NULL, 'n'}, -{"nsecs", no_argument, NULL, 'N'}, -{"oscope", required_argument, NULL, 'o'}, -{"priority", required_argument, NULL, 'p'}, - {"policy", required_argument, NULL, 'y'}, -{"preemptoff", no_argument, NULL, 'P'}, -{"quiet", no_argument, NULL, 'q'}, -{"relative", no_argument, NULL, 'r'}, -{"system", no_argument, NULL, 's'}, -{"threads", optional_argument, NULL, 't'}, -{"unbuffered", no_argument, NULL, 'u'}, -{"verbose", no_argument, NULL, 'v'}, -{"duration",required_argument, NULL, 'D'}, - {"wakeup", no_argument, NULL, 'w'}, - {"wakeuprt", no_argument, NULL, 'W'}, -{"help", no_argument, NULL, '?'}, -{"tracer", required_argument, NULL, 'T'}, -{"traceopt", required_argument, NULL, 'O'}, -{"smp", no_argument, NULL, 'S'}, -{"numa", no_argument, NULL, 'U'}, -{"latency", required_argument, NULL, 'e'}, -{"priospread", no_argument, NULL, 'Q'}, +{"affinity", optional_argument, NULL, 'a'}, +{"breaktrace", required_argument, NULL, 'b'}, +{"preemptirqs", no_argument, NULL, 'B'}, +{"clock", required_argument, NULL, 'c'}, +{"context", no_argument, NULL, 'C'}, +{"distance", required_argument, NULL, 'd'}, +{"duration", required_argument, NULL, 'D'}, +{"latency", required_argument, NULL, 'e'}, +{"event", no_argument, NULL, 'E'}, +{"ftrace", no_argument, NULL, 'f'}, +{"histogram", required_argument, NULL, 'h'}, +{"histofall", required_argument, NULL, 'H'}, +{"interval", required_argument, NULL, 'i'}, +{"irqsoff", no_argument, NULL, 'I'}, +{"loops", required_argument, NULL, 'l'}, +{"mlockall", no_argument, NULL, 'm'}, +{"refresh_on_max", no_argument, NULL, 'M'}, +{"nanosleep", no_argument, NULL, 'n'}, +{"nsecs", no_argument, NULL, 'N'}, +{"oscope", required_argument, NULL, 'o'}, +{"traceopt", required_argument, NULL, 'O'}, +{"priority", required_argument, NULL, 'p'}, +{"preemptoff", no_argument, NULL, 'P'}, +{"quiet", no_argument, NULL, 'q'}, +{"priospread", no_argument, NULL, 'Q'}, +{"relative", no_argument, NULL, 'r'}, +{"system", no_argument, NULL, 's'}, +{"smp", no_argument, NULL, 'S'}, +{"threads", optional_argument, NULL, 't'}, +{"tracer", required_argument, NULL, 'T'}, +{"unbuffered", no_argument, NULL, 'u'}, +{"numa", no_argument, NULL, 'U'}, +{"verbose", no_argument, NULL, 'v'}, +{"wakeup", no_argument, NULL, 'w'}, +{"wakeuprt", no_argument, NULL, 'W'}, +{"policy", required_argument, NULL, 'y'}, +{"help", no_argument, NULL, '?'}, {NULL, 0, NULL, 0} }; -int c = getopt_long(argc, argv, "a::b:Bc:Cd:Efh:H:i:Il:MnNo:O:p:PmqQrsSt::uUvD:wWT:y:e:", +int c = getopt_long(argc, argv, "a::b:Bc:Cd:D:e:Efh:H:i:Il:MnNo:O:p:PmqQrsSt::uUvD:wWT:y:", long_options, &amp;option_index); if (c == -1) break; &lt; at &gt;&lt; at &gt; -1109,6 +1113,13 &lt; at &gt;&lt; at &gt; static void process_options (int argc, c case 'c': clocksel = atoi(optarg); break; case 'C': tracetype = CTXTSWITCH; break; case 'd': distance = atoi(optarg); break; +case 'D': duration = parse_time_string(optarg); break; +case 'e': /* power management latency target value */ + /* note: default is 0 (zero) */ +latency_target_value = atoi(optarg); +if (latency_target_value &lt; 0) +latency_target_value = 0; +break; case 'E': enable_events = 1; break; case 'f': tracetype = FUNCTION; ftrace = 1; break; case 'H': histofall = 1; /* fall through */ &lt; at &gt;&lt; at &gt; -1124,6 +1135,8 &lt; at &gt;&lt; at &gt; static void process_options (int argc, c } break; case 'l': max_cycles = atoi(optarg); break; +case 'm': lockall = 1; break; +case 'M': refresh_on_max = 1; break; case 'n': use_nanosleep = MODE_CLOCK_NANOSLEEP; break; case 'N': use_nsecs = 1; break; case 'o': oscope_reduction = atoi(optarg); break; &lt; at &gt;&lt; at &gt; -1146,6 +1159,14 &lt; at &gt;&lt; at &gt; static void process_options (int argc, c case 'Q': priospread = 1; break; case 'r': timermode = TIMER_RELTIME; break; case 's': use_system = MODE_SYS_OFFSET; break; +case 'S': /* SMP testing */ +if (numa) +fatal("numa and smp options are mutually exclusive\n"); +smp = 1; +num_threads = max_cpus; +setaffinity = AFFINITY_USEALL; +use_nanosleep = MODE_CLOCK_NANOSLEEP; +break; case 't': if (smp) { warn("-t ignored due to --smp\n"); &lt; at &gt;&lt; at &gt; -1163,22 +1184,6 &lt; at &gt;&lt; at &gt; static void process_options (int argc, c strncpy(tracer, optarg, sizeof(tracer)); break; case 'u': setvbuf(stdout, NULL, _IONBF, 0); break; -case 'v': verbose = 1; break; -case 'm': lockall = 1; break; -case 'M': refresh_on_max = 1; break; -case 'D': duration = parse_time_string(optarg); -break; - case 'w': tracetype = WAKEUP; break; - case 'W': tracetype = WAKEUPRT; break; - case 'y': handlepolicy(optarg); break; -case 'S': /* SMP testing */ -if (numa) -fatal("numa and smp options are mutually exclusive\n"); -smp = 1; -num_threads = max_cpus; -setaffinity = AFFINITY_USEALL; -use_nanosleep = MODE_CLOCK_NANOSLEEP; -break; case 'U': /* NUMA testing */ if (smp) fatal("numa and smp options are mutually exclusive\n"); &lt; at &gt;&lt; at &gt; -1194,12 +1199,10 &lt; at &gt;&lt; at &gt; static void process_options (int argc, c warn("ignoring --numa or -U\n"); #endif break; -case 'e': /* power management latency target value */ - /* note: default is 0 (zero) */ -latency_target_value = atoi(optarg); -if (latency_target_value &lt; 0) -latency_target_value = 0; -break; +case 'v': verbose = 1; break; +case 'w': tracetype = WAKEUP; break; +case 'W': tracetype = WAKEUPRT; break; +case 'y': handlepolicy(optarg); break; case '?': display_help(0); break; } -- To unsubscribe from this list: send the line "unsubscribe linux-rt-users" in the body of a message to majordomo&lt; at &gt;vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html </pre> Frank Rowand 2012-05-15T23:36:08 http://comments.gmane.org/gmane.linux.rt.user/8296 <pre>Hello all, I was wondering what people used to check RT_PREEMPT behavior under load/stress? I'm trying to test the accuracy of my timers and have a test where I setup a kernel module with an hr-timer flipping RTS bit on serial COM0 periodically, which I can look on an oscilloscope. the scope triggers on rising edge, I call jitter what shows on the falling side: under no specific load I get ~ 10 us (worst case waiting a long time) My initial idea for stressing the system was to compile a kernel, make -j 8 (#cores) that I thought would exercise CPU and IO if anything. As it happens, it's "mostly good" but I do get occasional (but repeatable) wild excursions (&gt;100us) Looking around, I found a tool called 'stress' - http://weather.ou.edu/~apw/projects/stress/ Under these new conditions, the system behaves really well again ~20 us stable all the way. So both tests give different result, I'm not sure which to trust. I was thinking maybe there is some weird interaction with the kernel and building the kernel that make the 'bad' test invalid? I have RT_PREEMPT 3.0.18-rt34 SMP x86_64 Thanks, </pre> Matthieu Bec 2012-05-15T23:08:13 http://comments.gmane.org/gmane.linux.rt.user/8290 <pre>Trying to compile Preepmtion test with instructions from https://rt.wiki.kernel.org/index.php/Preemption_Test and I run into the error below g++ preempt-test.cc -g -lboost_thread-mt -lboost_program_options -o preempt-test preempt-test.cc:57:40: error: boost/thread/detail/lock.hpp: No such file or directory Is there a later version than v4 from ftp://ftp.novell.com/dev/ghaskins/preempt-test-latest.tar.bz2 The git link git://git.kernel.org/pub/scm/linux/kernel/git/ghaskins/preempt-test.git doesn't exist. Is there a git repo some where else. Thanks, Venkat -- To unsubscribe from this list: send the line "unsubscribe linux-rt-users" in the body of a message to majordomo&lt; at &gt;vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html </pre> Venkat Subbiah 2012-05-15T20:30:52 http://comments.gmane.org/gmane.linux.rt.user/8288 <pre>Ben, Found this while building v3.2.17-rt28: diff --git a/drivers/net/wireless/brcm80211/brcmsmac/main.c b/drivers/net/wireless/brcm80211/brcmsmac/main.c index f98becc..833cbef 100644 --- a/drivers/net/wireless/brcm80211/brcmsmac/main.c +++ b/drivers/net/wireless/brcm80211/brcmsmac/main.c &lt; at &gt;&lt; at &gt; -7879,7 +7879,7 &lt; at &gt;&lt; at &gt; brcms_c_recvctl(struct brcms_c_info *wlc, struct d11rxhdr *rxh, if (wlc-&gt;hw-&gt;suspended_fifos) { hdr = (struct ieee80211_hdr *)p-&gt;data; if (ieee80211_is_beacon(hdr-&gt;frame_control)) - brcms_b_mute(wlc-&gt;hw, false); + brcms_b_mute(wlc-&gt;hw, false, 0); } memcpy(IEEE80211_SKB_RXCB(p), &amp;rx_status, sizeof(rx_status)); Not sure if that's completely correct but I found another call site for brcms_b_mute() that also passed 0 flags. Clark </pre> Clark Williams 2012-05-15T19:04:13 http://comments.gmane.org/gmane.linux.rt.user/8276 <pre>The RT patch has been having lots of trouble lately with large machines and applications running lots of threads. This usually boils down to a bottle neck of a single lock: the mm-&gt;mmap_sem. The mmap_sem is a rwsem, which can sleep, but it also can be taken with a read/write lock, where a read lock can be taken by several tasks at the same time and the write lock can be only taken by a single task. But due to priority inheritance, having multiple readers makes the code much more complex, thus the -rt patch converts all rwsems into a single mutex, where readers may nest (the same task may grab the same rwsem for read multiple times), but only one task may hold the rwsem at any given time (for read or write). When we have lots of threads, the rwsem may be taken often, either for memory allocation or filling in page faults. This becomes a bottle neck for threads as only one thread at a time may grab the mmap_sem (which is shared by all threads of a process). Previous attempts of adding multiple readers became too complex and was error prone. This approach takes on a much more simpler technique, one that is actually used by per cpu locks. The idea here is to have an rwsem create a rt_mutex for each CPU. Actually, it creates a rwsem for each CPU that can only be acquired by one task at a time. This allows for readers on separate CPUs to take only the per cpu lock. When a writer needs to take a lock, it must grab all CPU locks before continuing. This approach does nothing special with the rt_mutex or priority inheritance code. That stays the same, and works normally (thus less error prone). The trick here is that when a reader takes a rwsem for read, it must disable migration, that way it can unlock the rwsem without needing any special searches (which lock did it take?). I've tested this a bit, and so far it works well. I haven't found a nice way to initialize the locks, so I'm using the silly initialize_rwsem() at all places that acquire the lock. But we can work on this later. Also, I don't use per_cpu sections for the locks, which means we have cache line collisions, but a normal (mainline) rwsem has that as well. These are all room for improvement (and why this is just an RFC patch). I'll see if I can get some numbers to see how this fixes the issues with multi threads on big boxes. Thoughts? </pre> Steven Rostedt 2012-05-15T14:03:45 http://comments.gmane.org/gmane.linux.rt.user/8274 <pre> Dear RT Folks, I'm pleased to announce the 3.2.17 stable release. This release is just an update to the new stable 3.2.17 version and no RT specific changes have been made. You can get this release via the git tree at: git://git.kernel.org/pub/scm/linux/kernel/git/rt/linux-stable-rt.git Head SHA1: 1b5c78bddb7cd30df391c91122d8d77114d8da37 Or to build 3.2.17 directly, the following patches should be applied: http://www.kernel.org/pub/linux/kernel/v3.x/linux-3.2.tar.xz http://www.kernel.org/pub/linux/kernel/v3.x/patch-3.2.17.xz http://www.kernel.org/pub/linux/kernel/projects/rt/3.2/patch-3.2.17.patch.xz Enjoy, </pre> Steven Rostedt 2012-05-14T18:09:04 http://comments.gmane.org/gmane.linux.rt.user/8273 <pre> Dear RT Folks, I'm pleased to announce the 3.0.31-rt51 stable release. This release is just an update to the new stable 3.0.31 version and no RT specific changes have been made. You can get this release via the git tree at: git://git.kernel.org/pub/scm/linux/kernel/git/rt/linux-stable-rt.git Head SHA1: f4162227ef21455c1b1bd74565e49e4c596cecb0 Or to build 3.0.31-rt51 directly, the following patches should be applied: http://www.kernel.org/pub/linux/kernel/v3.0/linux-3.0.tar.xz http://www.kernel.org/pub/linux/kernel/v3.0/patch-3.0.31.xz http://www.kernel.org/pub/linux/kernel/projects/rt/3.0/patch-3.0.31-rt51.patch.xz Enjoy, </pre> Steven Rostedt 2012-05-14T16:33:34 http://comments.gmane.org/gmane.linux.rt.user/8270 <pre>It's been a while since we've released a new rt-tests. Clark silently pushed v0.84 to git, so this note is to let you know. This release is git only - at least for now. Repo: git://git.kernel.org/pub/scm/linux/kernel/git/clrkwllms/rt-tests.git Branch: master Tag: v0.84 I also collect patches in my own repo to push to our fearless leader. :) This is at git://git.kernel.org/pub/scm/linux/kernel/git/jkacur/rt-tests.git Where I've started branch rt-tests-0.84-devel Thanks, enjoy! Clark Williams (5): [cyclictest] added priority spreading option --priospread Revert "cyclictest problem/bug as non-root" report number of samples written in hwlatdetect change cyclictest measurement thread to check returns and exit on error version bump to 0.84 Darren Hart (1): Makefile: Support user supplied CFLAGS and LDFLAGS Frank Rowand (6): rt-tests: cyclictest segfault with '-a' rt-tests: cyclictest avoid unneeded warning rt-tests: cyclictest warn of interaction between '-a', '--smp', and '--numa' rt-tests: Makefile get machinetype from compiler instead of uname rt-tests: incorrect first latency value for --verbose option rt-tests: printf format compile warning John Kacur (7): gitignore: differentiate between program names and directories pi_stress: Check the status of sched_getaffinity Makefile: Introduce a static library Move info, warn, and fatal functions to error.[ch] install: Fix failed to create symbolic link hwlatdetect file exists install: Fix failed to create symbolic link hwlatdetect file exists rt-tests / cyclictest: Make cyclictest fail if it cannot run with requested priority. Markus Kohlhase (1): added description from osadl.org Nicholas Mc Guire (1): cyclictest problem/bug as non-root Steven Rostedt (1): rt-tests: Update rt-migrate-test to use ftrace infrastructure -- To unsubscribe from this list: send the line "unsubscribe linux-rt-users" in the body of a message to majordomo&lt; at &gt;vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html </pre> John Kacur 2012-05-11T14:47:53 http://comments.gmane.org/gmane.linux.rt.user/8266 <pre> cyclictest getopt_long() parameter clean up. Clean up before following patch which will add a new option. Some elements of long_options were not in alphabetical order. Some elements of optstring were not in alphabetical order. '-e', '--latency' was missing help text short form of --duration ('D') was missing from optstring Change a few instances of leading spaces to tabs. Add white space to long_options to improve readability. Some cases of the switch processing the result of getopt_long() were not in alphabetical order. Did _not_ clean up option value parsing and processing. Signed-off-by: Frank Rowand &lt;frank.rowand&lt; at &gt;am.sony.com&gt; --- src/cyclictest/cyclictest.c | 143 73 +70 -0 ! 1 file changed, 73 insertions(+), 70 deletions(-) Index: b/src/cyclictest/cyclictest.c =================================================================== --- a/src/cyclictest/cyclictest.c +++ b/src/cyclictest/cyclictest.c &lt; at &gt;&lt; at &gt; -923,10 +923,11 &lt; at &gt;&lt; at &gt; static void display_help(int error) "-D --duration=t specify a length for the test run\n" " default is in seconds, but 'm', 'h', or 'd' maybe added\n" " to modify value to minutes, hours or days\n" + "-e --latency=PM_QOS write PM_QOS to /dev/cpu_dma_latency\n" "-E --event event tracing (used with -b)\n" "-f --ftrace function trace (when -b is active)\n" "-h --histogram=US dump a latency histogram to stdout after the run\n" - " (with same priority about many threads)\n" + " (with same priority about many threads)\n" " US is the max time to be be tracked in microseconds\n" "-H --histofall=US same as -h except with an additional summary column\n" "-i INTV --interval=INTV base interval of thread in us default=1000\n" &lt; at &gt;&lt; at &gt; -944,6 +945,8 &lt; at &gt;&lt; at &gt; static void display_help(int error) "-Q --priospread spread priority levels starting at specified value\n" "-r --relative use relative timer instead of absolute\n" "-s --system use sys_nanosleep and sys_setitimer\n" + "-S --smp Standard SMP testing: options -a -t -n and\n" + " same priority of all threads\n" "-t --threads one thread per available processor\n" "-t [NUM] --threads=NUM number of threads:\n" " without NUM, threads = max_cpus\n" &lt; at &gt;&lt; at &gt; -951,16 +954,14 &lt; at &gt;&lt; at &gt; static void display_help(int error) "-T TRACE --tracer=TRACER set tracing function\n" " configured tracers: %s\n" "-u --unbuffered force unbuffered output for live processing\n" + "-U --numa Standard NUMA testing (similar to SMP option)\n" + " thread data structures allocated from local node\n" "-v --verbose output values on stdout for statistics\n" " format: n:c:v n=tasknum c=count v=value in us\n" - "-w --wakeup task wakeup tracing (used with -b)\n" - "-W --wakeuprt rt task wakeup tracing (used with -b)\n" - "-y POLI --policy=POLI policy of realtime thread, POLI may be fifo(default) or rr\n" - " format: --policy=fifo(default) or --policy=rr\n" - "-S --smp Standard SMP testing: options -a -t -n and\n" - " same priority of all threads\n" - "-U --numa Standard NUMA testing (similar to SMP option)\n" - " thread data structures allocated from local node\n", + "-w --wakeup task wakeup tracing (used with -b)\n" + "-W --wakeuprt rt task wakeup tracing (used with -b)\n" + "-y POLI --policy=POLI policy of realtime thread, POLI may be fifo(default) or rr\n" + " format: --policy=fifo(default) or --policy=rr\n", tracers ); if (error) &lt; at &gt;&lt; at &gt; -1044,48 +1045,51 &lt; at &gt;&lt; at &gt; static void process_options (int argc, c for (;;) { int option_index = 0; -/** Options for getopt */ +/* + * Options for getopt + * Ordered alphabetically by single letter name + */ static struct option long_options[] = { -{"affinity", optional_argument, NULL, 'a'}, -{"breaktrace", required_argument, NULL, 'b'}, -{"preemptirqs", no_argument, NULL, 'B'}, -{"clock", required_argument, NULL, 'c'}, -{"context", no_argument, NULL, 'C'}, -{"distance", required_argument, NULL, 'd'}, -{"event", no_argument, NULL, 'E'}, -{"ftrace", no_argument, NULL, 'f'}, -{"histogram", required_argument, NULL, 'h'}, -{"histofall", required_argument, NULL, 'H'}, -{"interval", required_argument, NULL, 'i'}, -{"irqsoff", no_argument, NULL, 'I'}, -{"loops", required_argument, NULL, 'l'}, -{"mlockall", no_argument, NULL, 'm' }, -{"refresh_on_max", no_argument, NULL, 'M' }, -{"nanosleep", no_argument, NULL, 'n'}, -{"nsecs", no_argument, NULL, 'N'}, -{"oscope", required_argument, NULL, 'o'}, -{"priority", required_argument, NULL, 'p'}, - {"policy", required_argument, NULL, 'y'}, -{"preemptoff", no_argument, NULL, 'P'}, -{"quiet", no_argument, NULL, 'q'}, -{"relative", no_argument, NULL, 'r'}, -{"system", no_argument, NULL, 's'}, -{"threads", optional_argument, NULL, 't'}, -{"unbuffered", no_argument, NULL, 'u'}, -{"verbose", no_argument, NULL, 'v'}, -{"duration",required_argument, NULL, 'D'}, - {"wakeup", no_argument, NULL, 'w'}, - {"wakeuprt", no_argument, NULL, 'W'}, -{"help", no_argument, NULL, '?'}, -{"tracer", required_argument, NULL, 'T'}, -{"traceopt", required_argument, NULL, 'O'}, -{"smp", no_argument, NULL, 'S'}, -{"numa", no_argument, NULL, 'U'}, -{"latency", required_argument, NULL, 'e'}, -{"priospread", no_argument, NULL, 'Q'}, +{"affinity", optional_argument, NULL, 'a'}, +{"breaktrace", required_argument, NULL, 'b'}, +{"preemptirqs", no_argument, NULL, 'B'}, +{"clock", required_argument, NULL, 'c'}, +{"context", no_argument, NULL, 'C'}, +{"distance", required_argument, NULL, 'd'}, +{"duration", required_argument, NULL, 'D'}, +{"latency", required_argument, NULL, 'e'}, +{"event", no_argument, NULL, 'E'}, +{"ftrace", no_argument, NULL, 'f'}, +{"histogram", required_argument, NULL, 'h'}, +{"histofall", required_argument, NULL, 'H'}, +{"interval", required_argument, NULL, 'i'}, +{"irqsoff", no_argument, NULL, 'I'}, +{"loops", required_argument, NULL, 'l'}, +{"mlockall", no_argument, NULL, 'm'}, +{"refresh_on_max", no_argument, NULL, 'M'}, +{"nanosleep", no_argument, NULL, 'n'}, +{"nsecs", no_argument, NULL, 'N'}, +{"oscope", required_argument, NULL, 'o'}, +{"traceopt", required_argument, NULL, 'O'}, +{"priority", required_argument, NULL, 'p'}, +{"preemptoff", no_argument, NULL, 'P'}, +{"quiet", no_argument, NULL, 'q'}, +{"priospread", no_argument, NULL, 'Q'}, +{"relative", no_argument, NULL, 'r'}, +{"system", no_argument, NULL, 's'}, +{"smp", no_argument, NULL, 'S'}, +{"threads", optional_argument, NULL, 't'}, +{"tracer", required_argument, NULL, 'T'}, +{"unbuffered", no_argument, NULL, 'u'}, +{"numa", no_argument, NULL, 'U'}, +{"verbose", no_argument, NULL, 'v'}, +{"wakeup", no_argument, NULL, 'w'}, +{"wakeuprt", no_argument, NULL, 'W'}, +{"policy", required_argument, NULL, 'y'}, +{"help", no_argument, NULL, '?'}, {NULL, 0, NULL, 0} }; -int c = getopt_long(argc, argv, "a::b:Bc:Cd:Efh:H:i:Il:MnNo:O:p:PmqQrsSt::uUvD:wWT:y:e:", +int c = getopt_long(argc, argv, "a::b:Bc:Cd:D:e:Efh:H:i:Il:MnNo:O:p:PmqQrsSt::uUvD:wWT:y:", long_options, &amp;option_index); if (c == -1) break; &lt; at &gt;&lt; at &gt; -1109,6 +1113,13 &lt; at &gt;&lt; at &gt; static void process_options (int argc, c case 'c': clocksel = atoi(optarg); break; case 'C': tracetype = CTXTSWITCH; break; case 'd': distance = atoi(optarg); break; +case 'D': duration = parse_time_string(optarg); break; +case 'e': /* power management latency target value */ + /* note: default is 0 (zero) */ +latency_target_value = atoi(optarg); +if (latency_target_value &lt; 0) +latency_target_value = 0; +break; case 'E': enable_events = 1; break; case 'f': tracetype = FUNCTION; ftrace = 1; break; case 'H': histofall = 1; /* fall through */ &lt; at &gt;&lt; at &gt; -1124,6 +1135,8 &lt; at &gt;&lt; at &gt; static void process_options (int argc, c } break; case 'l': max_cycles = atoi(optarg); break; +case 'm': lockall = 1; break; +case 'M': refresh_on_max = 1; break; case 'n': use_nanosleep = MODE_CLOCK_NANOSLEEP; break; case 'N': use_nsecs = 1; break; case 'o': oscope_reduction = atoi(optarg); break; &lt; at &gt;&lt; at &gt; -1146,6 +1159,14 &lt; at &gt;&lt; at &gt; static void process_options (int argc, c case 'Q': priospread = 1; break; case 'r': timermode = TIMER_RELTIME; break; case 's': use_system = MODE_SYS_OFFSET; break; +case 'S': /* SMP testing */ +if (numa) +fatal("numa and smp options are mutually exclusive\n"); +smp = 1; +num_threads = max_cpus; +setaffinity = AFFINITY_USEALL; +use_nanosleep = MODE_CLOCK_NANOSLEEP; +break; case 't': if (smp) { warn("-t ignored due to --smp\n"); &lt; at &gt;&lt; at &gt; -1163,22 +1184,6 &lt; at &gt;&lt; at &gt; static void process_options (int argc, c strncpy(tracer, optarg, sizeof(tracer)); break; case 'u': setvbuf(stdout, NULL, _IONBF, 0); break; -case 'v': verbose = 1; break; -case 'm': lockall = 1; break; -case 'M': refresh_on_max = 1; break; -case 'D': duration = parse_time_string(optarg); -break; - case 'w': tracetype = WAKEUP; break; - case 'W': tracetype = WAKEUPRT; break; - case 'y': handlepolicy(optarg); break; -case 'S': /* SMP testing */ -if (numa) -fatal("numa and smp options are mutually exclusive\n"); -smp = 1; -num_threads = max_cpus; -setaffinity = AFFINITY_USEALL; -use_nanosleep = MODE_CLOCK_NANOSLEEP; -break; case 'U': /* NUMA testing */ if (smp) fatal("numa and smp options are mutually exclusive\n"); &lt; at &gt;&lt; at &gt; -1194,12 +1199,10 &lt; at &gt;&lt; at &gt; static void process_options (int argc, c warn("ignoring --numa or -U\n"); #endif break; -case 'e': /* power management latency target value */ - /* note: default is 0 (zero) */ -latency_target_value = atoi(optarg); -if (latency_target_value &lt; 0) -latency_target_value = 0; -break; +case 'v': verbose = 1; break; +case 'w': tracetype = WAKEUP; break; +case 'W': tracetype = WAKEUPRT; break; +case 'y': handlepolicy(optarg); break; case '?': display_help(0); break; } -- To unsubscribe from this list: send the line "unsubscribe linux-rt-users" in the body of a message to majordomo&lt; at &gt;vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html </pre> Frank Rowand 2012-05-11T00:59:40 http://comments.gmane.org/gmane.linux.rt.user/8265 <pre>I would like to know if it is possible to prioritize the Ethernet traffic till it reaches napi_gro_receive() or netif_receive_skb() called inside the network card driver. First I thought it is possible by the prioritization of the irq/XX-ethX thread which is used by the network card and the NET_RX_SOFTIRQ, NET_TX_SOFTIRQ threads. Till I read that the split from software interrupts to threads are gone in the RT patch. Now I'm a bit confused and unsure about the softirq handling. I used ftrace to determine the path of an incoming packet and it looks to me that the softirq is prozessed inside the irq/18-eth1 thread. So if I give the irq/18-eth1 a higher priority than the others, can I be sure that it gets not interrupted? here a part from my ftrace: irq/18-eth1-1738 [001] 1366.914708: function: e1000_intr &lt;-- irq_forced_thread_fn irq/18-eth1-1738 [001] 1366.914711: function: __napi_schedule &lt;-- e1000_intr [...] irq/18-eth1-1738 [001] 1366.914718: function: __do_softirq &lt;-- local_bh_enable irq/18-eth1-1738 [001] 1366.914719: function: __do_softirq_common &lt;-- __do_softirq irq/18-eth1-1738 [001] 1366.914719: function: net_rx_action &lt;-- __do_softirq_common irq/18-eth1-1738 [001] 1366.914720: function: e1000_clean &lt;-- net_rx_action irq/18-eth1-1738 [001] 1366.914721: function: e1000_clean_rx_irq &lt;-- e1000_clean irq/18-eth1-1738 [001] 1366.914722: function: __netdev_alloc_skb &lt;-- e1000_clean_rx_irq irq/18-eth1-1738 [001] 1366.914722: function: __alloc_skb &lt;-- __netdev_alloc_skb [...] irq/18-eth1-1738 [001] 1366.914744: function: e1000_receive_skb &lt;-- e1000_clean_rx_irq irq/18-eth1-1738 [001] 1366.914744: function: eth_type_trans &lt;-- e1000_receive_skb irq/18-eth1-1738 [001] 1366.914744: function: napi_gro_receive &lt;-- e1000_receive_skb I hope someone can explain it to me. The reason I want to prioritize the incoming network traffic is cause I'm writing a module for the TTEthernet protocol which is Time-triggered and and based on the standard Ethernet. I need the packets without or very less jitter. And I would like to use it by changing the napi_gro_receive() funtion in the network card driver to my module function to use it with any card by almost no modification in the code. I'm quite new to kernel development and the module I'm writing is part of my bachelor thesis. If there is a paper or a book who explains this behavior I'm happy to know about it! Best regards Frieder -- To unsubscribe from this list: send the line "unsubscribe linux-rt-users" in the body of a message to majordomo&lt; at &gt;vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html </pre> Frieder Rick 2012-05-10T21:42:46 http://comments.gmane.org/gmane.linux.rt.user/8256 <pre> I've been doing some interrupt latency testing using a board running a 400MHz ARM9 (an Atmel at19sam9g20). I'm measuring interrupt latency of an ISR attached to external interrupt pin IRQ0. I built three kernels from scratch and ran the same test on all three: 2.6.30 2.6.33.7 2.6.33.7-rt30 All kernels are built using the standard AT91 patches from http://maxim.org.za/at91_26.html. The 2.6.30 kernel also has a set of patches distributed by Atmel. The 2.6.33.7-rt30 kernel is built from the same sources as the 2.6.33.7 kernel with the addition of the 2.6.33.6-rt30 patch. I don't have enough data to say much about worst-case latency, but after looking a few thousand samples I can say that... * 2.6.30 and 2.6.33.7 are pretty much the same: there is as much variation between test runs as there is between kernels. * Typical latency with RT patch is 3X the latency without it. * CPU idle time during test is 10% with RT patch and 30% without. IOW: the real-time patch for 2.6.33.7 makes both the typical interrupt latency and the CPU usage significantly worse. Typical latency without the RT patch is 5-15us. Typical latency with is 15-50us (I've never seen latency below 15us with the RT patch). Is this the expected behavior? I don't have enough data to state conclusively what the worst-case latencies are, but it looks like in all cases they're 200-250us. In all cases the kernel was configured to be the most preemptible it could be. In the 2.6.33.7-rt30 kernel I don't have any of the tracers enabled. The kernel configurations are as identical as I could make them. In the 2.6.33.7-rt30 case, I started with the .config file used to build the other 2.6.33.7 kernel and ran old-config making sure to select the "fully preemptible" option when it asked. Why would a "vanilla" 2.6.33.7 AT91 kernel perform so much better than one with the RT patch? </pre> Grant Edwards 2012-05-09T21:18:10 http://comments.gmane.org/gmane.linux.rt.user/8254 <pre>Minor fix to make working with git nicer. Now that we're building a lib, we need to exclude it from git status output. Do the same for patches we generate or apply. Signed-off-by: John Kacur &lt;jkacur&lt; at &gt;redhat.com&gt; --- .gitignore | 2 ++ 1 files changed, 2 insertions(+), 0 deletions(-) diff --git a/.gitignore b/.gitignore index 047925c..1924e73 100644 --- a/.gitignore +++ b/.gitignore &lt; at &gt;&lt; at &gt; -3,6 +3,8 &lt; at &gt;&lt; at &gt; *.o *.tar.gz *.d +*.patch +*.a tmp patches ChangeLog </pre> John Kacur 2012-05-09T15:37:36 http://comments.gmane.org/gmane.linux.rt.user/8253 <pre>Hi everyone, I am facing a problem while trying to run a rt-patched 3.2.y kernel with PREEMPT_RT_FULL on a mini2440 board (ARM based s3c2440 SoC) : whilst a vanilla kernel (with PREEMPT_LL) works like a charm, PREEMPT_RTB and PREEMPT_RT_FULL makes the MMC/SD driver (s3cmci) hang, which result in the kernel waiting indefinitely for the root fs to mount. I think that the IRQ handling of the driver is somehow disturbed by the changes made by the RT patch. When using PREEMPT_RTB I can see the following message in the console : BUG: scheduling while atomic: irq/37-s3c-mci/253/0x00000102 Modules linked in: Function entered at [&lt;c000e90c&gt;] from [&lt;c029f478&gt;] Function entered at [&lt;c029f478&gt;] from [&lt;c029fc3c&gt;] Function entered at [&lt;c029fc3c&gt;] from [&lt;c00559c0&gt;] Function entered at [&lt;c00559c0&gt;] from [&lt;c01dd4c8&gt;] Function entered at [&lt;c01dd4c8&gt;] from [&lt;c0024a0c&gt;] Function entered at [&lt;c0024a0c&gt;] from [&lt;c0024d9c&gt;] Function entered at [&lt;c0024d9c&gt;] from [&lt;c0024fe8&gt;] Function entered at [&lt;c0024fe8&gt;] from [&lt;c0025140&gt;] Function entered at [&lt;c0025140&gt;] from [&lt;c00549f0&gt;] Function entered at [&lt;c00549f0&gt;] from [&lt;c00547c4&gt;] Function entered at [&lt;c00547c4&gt;] from [&lt;c0039954&gt;] Function entered at [&lt;c0039954&gt;] from [&lt;c000a120&gt;] When I run the kernel under Qemu, debug through gdb and put a breakpoint on unwind_backtace the details of the previous backtrace is : #0 unwind_backtrace (regs=0x0, tsk=0x0) at arch/arm/kernel/unwind.c:409 #1 0xc029f478 in schedule_debug (prev=&lt;optimized out&gt;) at kernel/sched.c:4357 #2 __schedule () at kernel/sched.c:4537 #3 0xc029fc3c in schedule () at kernel/sched.c:4625 #4 0xc00559c0 in synchronize_irq (irq=&lt;optimized out&gt;) at kernel/irq/manage.c:73 Backtrace stopped: previous frame inner to this frame (corrupt stack?) I don't see the "bug" message with PREEMPT_RT_FULL, the kernel just hang waiting for the rootfs. The problem did not occur in the 2.6.y tree AFAIK. I'd like to investigate more and fix the s3cmci driver but I don't know where to begin, could someone give me some hints on what should be done/not be done for a rt patch compliant device driver/irq handler ? Regards, Christophe Huriaux -- To unsubscribe from this list: send the line "unsubscribe linux-rt-users" in the body of a message to majordomo&lt; at &gt;vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html </pre> Christophe Huriaux 2012-05-09T14:07:13 http://comments.gmane.org/gmane.linux.rt.user/8252 <pre>Currently if a non-root user requests a priority higher than the soft limit in /etc/security/limits.conf the call to sched_setscheduler will silently fail and the user will be running with priority of 0. Cyclictest will not complain, and display the requested priority resulting in seemingly poor results. The following patch fixes this by doing two things. 1. If the requested priority is higher than the soft limit but lower than the hard limit, it will raise the soft limit to the requested priority. 2. If the requested priority is higher than the hard limit, it will fail with a warning. The patch should not affect privileged users. Reported-by: Nicholas Mc Guire &lt;der.herr&lt; at &gt;hofr.at&gt; Signed-off-by: John Kacur &lt;jkacur&lt; at &gt;redhat.com&gt; --- src/cyclictest/cyclictest.c | 80 ++++++++++++++++++++++++++++++++++++++++++- 1 files changed, 79 insertions(+), 1 deletions(-) diff --git a/src/cyclictest/cyclictest.c b/src/cyclictest/cyclictest.c index e8649c7..c44df6a 100644 --- a/src/cyclictest/cyclictest.c +++ b/src/cyclictest/cyclictest.c &lt; at &gt;&lt; at &gt; -12,6 +12,7 &lt; at &gt;&lt; at &gt; #include &lt;stdio.h&gt; #include &lt;stdlib.h&gt; + #include &lt;stdint.h&gt; #include &lt;stdarg.h&gt; #include &lt;unistd.h&gt; &lt; at &gt;&lt; at &gt; -31,6 +32,7 &lt; at &gt;&lt; at &gt; #include &lt;sys/sysinfo.h&gt; #include &lt;sys/types.h&gt; #include &lt;sys/time.h&gt; +#include &lt;sys/resource.h&gt; #include &lt;sys/utsname.h&gt; #include &lt;sys/mman.h&gt; #include "rt_numa.h" &lt; at &gt;&lt; at &gt; -586,6 +588,79 &lt; at &gt;&lt; at &gt; parse_time_string(char *val) } /* + * Raise the soft priority limit up to prio, if that is less than or equal + * to the hard limit + * if a call fails, return the error + * if successful return 0 + * if fails, return -1 +*/ +static int raise_soft_prio(int policy, const struct sched_param *param) +{ +int err; +int policy_max;/* max for scheduling policy such as SCHED_FIFO */ +int soft_max; +int hard_max; +int prio; +struct rlimit rlim; + +prio = param-&gt;sched_priority; + +policy_max = sched_get_priority_max(policy); +if (policy_max == -1) { +err = errno; +err_msg("WARN: no such policy\n"); +return err; +} + +err = getrlimit(RLIMIT_RTPRIO, &amp;rlim); +if (err) { +err = errno; +err_msg_n(err, "WARN: getrlimit failed\n"); +return err; +} + +soft_max = (rlim.rlim_cur == RLIM_INFINITY) ? policy_max : rlim.rlim_cur; +hard_max = (rlim.rlim_max == RLIM_INFINITY) ? policy_max : rlim.rlim_max; + +if (prio &gt; soft_max &amp;&amp; prio &lt;= hard_max) { +rlim.rlim_cur = prio; +err = setrlimit(RLIMIT_RTPRIO, &amp;rlim); +if (err) { +err = errno; +err_msg_n(err, "WARN: setrlimit failed\n"); +/* return err; */ +} +} else { +err = -1; +} + +return err; +} + +/* + * Check the error status of sched_setscheduler + * If an error can be corrected by raising the soft limit priority to + * a priority less than or equal to the hard limit, then do so. + */ +static int setscheduler(pid_t pid, int policy, const struct sched_param *param) +{ +int err = 0; + +try_again: +err = sched_setscheduler(pid, policy, param); +if (err) { +err = errno; +if (err == EPERM) { +int err1; +err1 = raise_soft_prio(policy, param); +if (!err1) goto try_again; +} +} + +return err; +} + +/* * timer thread * * Modes: &lt; at &gt;&lt; at &gt; -599,6 +674,7 &lt; at &gt;&lt; at &gt; parse_time_string(char *val) */ void *timerthread(void *param) { +int err; struct thread_param *par = param; struct sched_param schedp; struct sigevent sigev; &lt; at &gt;&lt; at &gt; -643,7 +719,9 &lt; at &gt;&lt; at &gt; void *timerthread(void *param) memset(&amp;schedp, 0, sizeof(schedp)); schedp.sched_priority = par-&gt;prio; -sched_setscheduler(0, par-&gt;policy, &amp;schedp); +err = setscheduler(0, par-&gt;policy, &amp;schedp); +if (err) +fatal("Couldn't raise priority to %d\n", schedp.sched_priority); /* Get current time */ clock_gettime(par-&gt;clock, &amp;now); </pre> John Kacur 2012-05-09T14:03:03 Search the mailing list at Gmane query http://search.gmane.org/?group=$group=gmane.linux.rt.user