dd: Improve output format

Include written bytes and speed with units

BAN/include/BAN/CircularQueue.h

@@ -1,6 +1,7 @@
 #pragma once
 
 #include <BAN/Assert.h>
+#include <BAN/PlacementNew.h>
 
 #include <stdint.h>
 #include <stddef.h>

BAN/include/BAN/Function.h

@@ -2,7 +2,7 @@
 
 #include <BAN/Errors.h>
 #include <BAN/Move.h>
-#include <BAN/New.h>
+#include <BAN/PlacementNew.h>
 
 namespace BAN
 {

BAN/include/BAN/LinkedList.h

@@ -3,6 +3,7 @@
 #include <BAN/Errors.h>
 #include <BAN/Move.h>
 #include <BAN/New.h>
+#include <BAN/PlacementNew.h>
 
 namespace BAN
 {

BAN/include/BAN/New.h

@@ -16,6 +16,3 @@ namespace BAN
 		static constexpr void(&deallocator)(void*) = free;
 	#endif
 }
-
-inline void* operator new(size_t, void* addr)	{ return addr; }
-inline void* operator new[](size_t, void* addr)	{ return addr; }

BAN/include/BAN/Optional.h

@@ -2,6 +2,7 @@
 
 #include <BAN/Assert.h>
 #include <BAN/Move.h>
+#include <BAN/PlacementNew.h>
 
 #include <stdint.h>
 

BAN/include/BAN/PlacementNew.h

@@ -0,0 +1,4 @@
+#pragma once
+
+inline void* operator new(size_t, void* addr)	{ return addr; }
+inline void* operator new[](size_t, void* addr)	{ return addr; }

BAN/include/BAN/Queue.h

@@ -5,6 +5,7 @@
 #include <BAN/Math.h>
 #include <BAN/Move.h>
 #include <BAN/New.h>
+#include <BAN/PlacementNew.h>
 
 namespace BAN
 {

BAN/include/BAN/Variant.h

@@ -3,7 +3,7 @@
 #include <BAN/Assert.h>
 #include <BAN/Math.h>
 #include <BAN/Move.h>
-#include <BAN/New.h>
+#include <BAN/PlacementNew.h>
 
 #include <string.h>
 

BAN/include/BAN/Vector.h

@@ -5,6 +5,7 @@
 #include <BAN/Math.h>
 #include <BAN/Move.h>
 #include <BAN/New.h>
+#include <BAN/PlacementNew.h>
 #include <BAN/Span.h>
 
 namespace BAN

kernel/arch/x86_64/linker.ld

@@ -19,11 +19,11 @@ SECTIONS
 		g_userspace_start = .;
 		*(.userspace)
 		g_userspace_end = .;
+		g_kernel_execute_end = .;
 	}
 	.rodata ALIGN(4K) : AT(ADDR(.rodata) - KERNEL_OFFSET)
 	{
 		*(.rodata.*)
-		g_kernel_execute_end = .;
 	}
 	.data ALIGN(4K) : AT(ADDR(.data) - KERNEL_OFFSET)
 	{

kernel/include/kernel/SpinLock.h

@@ -37,7 +37,6 @@ namespace Kernel
 	private:
 		BAN::Atomic<pid_t>		m_locker		= -1;
 		BAN::Atomic<uint32_t>	m_lock_depth	= 0;
-		SpinLock m_lock;
 	};
 
 	class RecursivePrioritySpinLock
@@ -54,10 +53,9 @@ namespace Kernel
 		uint32_t lock_depth() const { return m_lock_depth; }
 
 	private:
-		pid_t m_locker = -1;
+		BAN::Atomic<pid_t>		m_locker		= -1;
-		uint32_t m_queue_length = 0;
+		BAN::Atomic<uint32_t>	m_lock_depth	= 0;
-		uint32_t m_lock_depth = 0;
+		BAN::Atomic<uint32_t>	m_queue_length	= 0;
-		SpinLock m_lock;
 	};
 
 }

kernel/kernel/SpinLock.cpp

@@ -25,40 +25,21 @@ namespace Kernel
 	void RecursiveSpinLock::lock()
 	{
 		pid_t tid = Scheduler::current_tid();
-
+		if (m_locker != tid)
-		while (true)
-		{
-			m_lock.lock();
-			if (m_locker == tid)
 		{
+			while (!m_locker.compare_exchange(-1, tid))
+				Scheduler::get().reschedule();
+			ASSERT(m_lock_depth == 0);
+		}
 		m_lock_depth++;
-				break;
-			}
-			if (m_locker == -1)
-			{
-				m_locker = tid;
-				m_lock_depth = 1;
-				break;
-			}
-			m_lock.unlock();
-		}
-
-		m_lock.unlock();
 	}
 
 	void RecursiveSpinLock::unlock()
 	{
-		m_lock.lock();
-
 		ASSERT(m_lock_depth > 0);
 		ASSERT(m_locker == Scheduler::current_tid());
-
+		if (--m_lock_depth == 0)
-		m_lock_depth--;
-
-		if (m_lock_depth == 0)
 			m_locker = -1;
-
-		m_lock.unlock();
 	}
 
 	bool RecursiveSpinLock::is_locked() const
@@ -71,38 +52,20 @@ namespace Kernel
 		pid_t tid = Scheduler::current_tid();
 
 		bool has_priority = !Thread::current().is_userspace();
-
 		if (has_priority)
-		{
-			m_lock.lock();
 			m_queue_length++;
-			m_lock.unlock();
-		}
 
-		while (true)
+		if (m_locker != tid)
-		{
-			m_lock.lock();
-			if (m_locker == tid)
 		{
+			while (!((has_priority || m_queue_length == 0) && m_locker.compare_exchange(-1, tid)))
+				Scheduler::get().reschedule();
+			ASSERT(m_lock_depth == 0);
+		}
 		m_lock_depth++;
-				break;
-			}
-			if (m_locker == -1 && (has_priority || m_queue_length == 0))
-			{
-				m_locker = tid;
-				m_lock_depth = 1;
-				break;
-			}
-			m_lock.unlock();
-		}
-
-		m_lock.unlock();
 	}
 
 	void RecursivePrioritySpinLock::unlock()
 	{
-		m_lock.lock();
-
 		ASSERT(m_lock_depth > 0);
 		ASSERT(m_locker == Scheduler::current_tid());
 
@@ -110,12 +73,8 @@ namespace Kernel
 		if (has_priority)
 			m_queue_length--;
 
-		m_lock_depth--;
+		if (--m_lock_depth == 0)
-
-		if (m_lock_depth == 0)
 			m_locker = -1;
-
-		m_lock.unlock();
 	}
 
 	bool RecursivePrioritySpinLock::is_locked() const

userspace/dd/main.cpp

@@ -1,5 +1,6 @@
 #include <ctype.h>
 #include <fcntl.h>
+#include <inttypes.h>
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
@@ -7,35 +8,91 @@
 
 #define CURRENT_NS() ({ timespec ts; clock_gettime(CLOCK_MONOTONIC, &ts); ts.tv_sec * 1'000'000'000 + ts.tv_nsec; })
 
-int parse_int(const char* val)
+uint64_t parse_sized_u64(const char* val)
 {
-	int result = 0;
+	uint64_t result = 0;
-	for (const char* ptr = val; *ptr; ptr++)
+
+	const char* ptr = val;
+	for (; *ptr && isdigit(*ptr); ptr++)
+		result = (result * 10) + (*ptr - '0');
+
+	switch (*ptr)
 	{
-		if (!isdigit(*ptr))
+		case 'E': result *= 1024; // fall through
+		case 'P': result *= 1024; // fall through
+		case 'T': result *= 1024; // fall through
+		case 'G': result *= 1024; // fall through
+		case 'M': result *= 1024; // fall through
+		case 'K': case 'k': result *= 1024; ptr++; break;
+	}
+
+	if (*ptr != '\0')
 	{
 		fprintf(stderr, "invalid number: %s\n", val);
 		exit(1);
 	}
-		result = (result * 10) + (*ptr - '0');
+
-	}
 	return result;
 }
 
-void print_time(uint64_t start_ns, uint64_t end_ns, size_t transfered)
+void print_value_with_unit(uint64_t value_x10, unsigned base, const char* units[])
+{
+	unsigned index = 0;
+	while (value_x10 / 10 >= base)
+	{
+		index++;
+		value_x10 /= base;
+	}
+	if (value_x10 < 100)
+		printf("%u.%u %s", (unsigned)value_x10 / 10, (unsigned)value_x10 % 10, units[index]);
+	else
+		printf("%u %s", (unsigned)value_x10 / 10, units[index]);
+
+}
+
+void print_time(uint64_t start_ns, uint64_t end_ns, uint64_t transfered, bool last = false)
 {
 	static bool first = true;
-	uint64_t duration_ns = end_ns - start_ns;
+	if (!first)
-	printf("%s%zu bytes copied, %d.%09d s\e[K\n", (first ? "" : "\e[F"), transfered, (int)(duration_ns / 1'000'000'000), (int)(duration_ns % 1'000'000'000));
+		printf("\e[F");
 	first = false;
+
+	printf("%" PRIu64 " bytes", transfered);
+	if (transfered >= 1000)
+	{
+		printf(" (");
+		{
+			const char* units[] { "", "kB", "MB", "GB", "TB", "PB", "EB", "ZB", "YB", "RB", "QB" };
+			print_value_with_unit(transfered * 10, 1000, units);
+		}
+		if (transfered >= 1024)
+		{
+			printf(", ");
+			const char* units[] { "", "KiB", "MiB", "GiB", "TiB", "PiB", "EiB", "ZiB", "YiB", "RiB", "QiB" };
+			print_value_with_unit(transfered * 10, 1024, units);
+		}
+		printf(")");
+	}
+	printf(" copied");
+
+	double duration_s = (end_ns - start_ns) / 1e9;
+	if (last)
+		printf(", %f s, ", duration_s);
+	else
+		printf(", %u s, ", (unsigned)duration_s);
+
+	const char* units[] { "B/s", "kB/s", "MB/s", "GB/s", "TB/s", "PB/s", "EB/s", "ZB/s", "YB/s", "RB/s", "QB/s" };
+	print_value_with_unit(10 * transfered / duration_s, 1000, units);
+
+	printf("\e[K\n");
 }
 
 int main(int argc, char** argv)
 {
 	const char* input = nullptr;
 	const char* output = nullptr;
-	int bs = 512;
+	uint64_t bs = 512;
-	int count = -1;
+	uint64_t count = ~(uint64_t)0;
 	bool print_progress = false;
 
 	for (int i = 1; i < argc; i++)
@@ -45,9 +102,9 @@ int main(int argc, char** argv)
 		else if (strncmp(argv[i], "of=", 3) == 0)
 			output = argv[i] + 3;
 		else if (strncmp(argv[i], "bs=", 3) == 0)
-			bs = parse_int(argv[i] + 3);
+			bs = parse_sized_u64(argv[i] + 3);
 		else if (strncmp(argv[i], "count=", 6) == 0)
-			count = parse_int(argv[i] + 6);
+			count = parse_sized_u64(argv[i] + 6);
 		else if (strcmp(argv[i], "status=progress") == 0)
 			print_progress = true;
 		else
@@ -86,12 +143,11 @@ int main(int argc, char** argv)
 		return 1;
 	}
 
-	size_t total_transfered = 0;
+	uint64_t total_transfered = 0;
-
 	uint64_t start_ns = CURRENT_NS();
-	uint64_t last_print_ns = 0;
+	uint64_t last_print_s = 0;
 
-	for (int i = 0; i != count; i++)
+	for (uint64_t i = 0; i != count; i++)
 	{
 		ssize_t nread = read(ifd, buffer, bs);
 		if (nread == -1)
@@ -109,21 +165,21 @@ int main(int argc, char** argv)
 
 		total_transfered += nwrite;
 
-		if (nread < bs || nwrite < bs)
+		if ((size_t)nread < bs || (size_t)nwrite < bs)
 			break;
 
 		if (print_progress)
 		{
 			uint64_t current_ns = CURRENT_NS();
-			if (current_ns >= last_print_ns + 1'000'000'000)
+			if ((current_ns - start_ns) / 1'000'000'000 > last_print_s)
 			{
 				print_time(start_ns, current_ns, total_transfered);
-				last_print_ns = current_ns;
+				last_print_s = (current_ns - start_ns) / 1'000'000'000;
 			}
 		}
 	}
 
-	print_time(start_ns, CURRENT_NS(), total_transfered);
+	print_time(start_ns, CURRENT_NS(), total_transfered, true);
 
 	close(ifd);
 	close(ofd);