linux/arch/blackfin/mm/isram-driver.c

411 lines
10 KiB
C
Raw Normal View History

/*
* Instruction SRAM accessor functions for the Blackfin
*
* Copyright 2008 Analog Devices Inc.
*
* Licensed under the GPL-2 or later
*/
#define pr_fmt(fmt) "isram: " fmt
#include <linux/module.h>
#include <linux/kernel.h>
#include <linux/types.h>
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h percpu.h is included by sched.h and module.h and thus ends up being included when building most .c files. percpu.h includes slab.h which in turn includes gfp.h making everything defined by the two files universally available and complicating inclusion dependencies. percpu.h -> slab.h dependency is about to be removed. Prepare for this change by updating users of gfp and slab facilities include those headers directly instead of assuming availability. As this conversion needs to touch large number of source files, the following script is used as the basis of conversion. http://userweb.kernel.org/~tj/misc/slabh-sweep.py The script does the followings. * Scan files for gfp and slab usages and update includes such that only the necessary includes are there. ie. if only gfp is used, gfp.h, if slab is used, slab.h. * When the script inserts a new include, it looks at the include blocks and try to put the new include such that its order conforms to its surrounding. It's put in the include block which contains core kernel includes, in the same order that the rest are ordered - alphabetical, Christmas tree, rev-Xmas-tree or at the end if there doesn't seem to be any matching order. * If the script can't find a place to put a new include (mostly because the file doesn't have fitting include block), it prints out an error message indicating which .h file needs to be added to the file. The conversion was done in the following steps. 1. The initial automatic conversion of all .c files updated slightly over 4000 files, deleting around 700 includes and adding ~480 gfp.h and ~3000 slab.h inclusions. The script emitted errors for ~400 files. 2. Each error was manually checked. Some didn't need the inclusion, some needed manual addition while adding it to implementation .h or embedding .c file was more appropriate for others. This step added inclusions to around 150 files. 3. The script was run again and the output was compared to the edits from #2 to make sure no file was left behind. 4. Several build tests were done and a couple of problems were fixed. e.g. lib/decompress_*.c used malloc/free() wrappers around slab APIs requiring slab.h to be added manually. 5. The script was run on all .h files but without automatically editing them as sprinkling gfp.h and slab.h inclusions around .h files could easily lead to inclusion dependency hell. Most gfp.h inclusion directives were ignored as stuff from gfp.h was usually wildly available and often used in preprocessor macros. Each slab.h inclusion directive was examined and added manually as necessary. 6. percpu.h was updated not to include slab.h. 7. Build test were done on the following configurations and failures were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my distributed build env didn't work with gcov compiles) and a few more options had to be turned off depending on archs to make things build (like ipr on powerpc/64 which failed due to missing writeq). * x86 and x86_64 UP and SMP allmodconfig and a custom test config. * powerpc and powerpc64 SMP allmodconfig * sparc and sparc64 SMP allmodconfig * ia64 SMP allmodconfig * s390 SMP allmodconfig * alpha SMP allmodconfig * um on x86_64 SMP allmodconfig 8. percpu.h modifications were reverted so that it could be applied as a separate patch and serve as bisection point. Given the fact that I had only a couple of failures from tests on step 6, I'm fairly confident about the coverage of this conversion patch. If there is a breakage, it's likely to be something in one of the arch headers which should be easily discoverable easily on most builds of the specific arch. Signed-off-by: Tejun Heo <tj@kernel.org> Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org> Cc: Ingo Molnar <mingo@redhat.com> Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 16:04:11 +08:00
#include <linux/slab.h>
#include <linux/spinlock.h>
#include <linux/sched.h>
#include <asm/blackfin.h>
#include <asm/dma.h>
/*
* IMPORTANT WARNING ABOUT THESE FUNCTIONS
*
* The emulator will not function correctly if a write command is left in
* ITEST_COMMAND or DTEST_COMMAND AND access to cache memory is needed by
* the emulator. To avoid such problems, ensure that both ITEST_COMMAND
* and DTEST_COMMAND are zero when exiting these functions.
*/
/*
* On the Blackfin, L1 instruction sram (which operates at core speeds) can not
* be accessed by a normal core load, so we need to go through a few hoops to
* read/write it.
* To try to make it easier - we export a memcpy interface, where either src or
* dest can be in this special L1 memory area.
* The low level read/write functions should not be exposed to the rest of the
* kernel, since they operate on 64-bit data, and need specific address alignment
*/
static DEFINE_SPINLOCK(dtest_lock);
/* Takes a void pointer */
#define IADDR2DTEST(x) \
({ unsigned long __addr = (unsigned long)(x); \
((__addr & (1 << 11)) << (26 - 11)) | /* addr bit 11 (Way0/Way1) */ \
(1 << 24) | /* instruction access = 1 */ \
((__addr & (1 << 15)) << (23 - 15)) | /* addr bit 15 (Data Bank) */ \
((__addr & (3 << 12)) << (16 - 12)) | /* addr bits 13:12 (Subbank) */ \
(__addr & 0x47F8) | /* addr bits 14 & 10:3 */ \
(1 << 2); /* data array = 1 */ \
})
/* Takes a pointer, and returns the offset (in bits) which things should be shifted */
#define ADDR2OFFSET(x) ((((unsigned long)(x)) & 0x7) * 8)
/* Takes a pointer, determines if it is the last byte in the isram 64-bit data type */
#define ADDR2LAST(x) ((((unsigned long)x) & 0x7) == 0x7)
static void isram_write(const void *addr, uint64_t data)
{
uint32_t cmd;
unsigned long flags;
if (unlikely(addr >= (void *)(L1_CODE_START + L1_CODE_LENGTH)))
return;
cmd = IADDR2DTEST(addr) | 2; /* write */
/*
* Writes to DTEST_DATA[0:1] need to be atomic with write to DTEST_COMMAND
* While in exception context - atomicity is guaranteed or double fault
*/
spin_lock_irqsave(&dtest_lock, flags);
bfin_write_DTEST_DATA0(data & 0xFFFFFFFF);
bfin_write_DTEST_DATA1(data >> 32);
/* use the builtin, since interrupts are already turned off */
__builtin_bfin_csync();
bfin_write_DTEST_COMMAND(cmd);
__builtin_bfin_csync();
bfin_write_DTEST_COMMAND(0);
__builtin_bfin_csync();
spin_unlock_irqrestore(&dtest_lock, flags);
}
static uint64_t isram_read(const void *addr)
{
uint32_t cmd;
unsigned long flags;
uint64_t ret;
if (unlikely(addr > (void *)(L1_CODE_START + L1_CODE_LENGTH)))
return 0;
cmd = IADDR2DTEST(addr) | 0; /* read */
/*
* Reads of DTEST_DATA[0:1] need to be atomic with write to DTEST_COMMAND
* While in exception context - atomicity is guaranteed or double fault
*/
spin_lock_irqsave(&dtest_lock, flags);
/* use the builtin, since interrupts are already turned off */
__builtin_bfin_csync();
bfin_write_DTEST_COMMAND(cmd);
__builtin_bfin_csync();
ret = bfin_read_DTEST_DATA0() | ((uint64_t)bfin_read_DTEST_DATA1() << 32);
bfin_write_DTEST_COMMAND(0);
__builtin_bfin_csync();
spin_unlock_irqrestore(&dtest_lock, flags);
return ret;
}
static bool isram_check_addr(const void *addr, size_t n)
{
if ((addr >= (void *)L1_CODE_START) &&
(addr < (void *)(L1_CODE_START + L1_CODE_LENGTH))) {
if (unlikely((addr + n) > (void *)(L1_CODE_START + L1_CODE_LENGTH))) {
show_stack(NULL, NULL);
pr_err("copy involving %p length (%zu) too long\n", addr, n);
}
return true;
}
return false;
}
/*
* The isram_memcpy() function copies n bytes from memory area src to memory area dest.
* The isram_memcpy() function returns a pointer to dest.
* Either dest or src can be in L1 instruction sram.
*/
void *isram_memcpy(void *dest, const void *src, size_t n)
{
uint64_t data_in = 0, data_out = 0;
size_t count;
bool dest_in_l1, src_in_l1, need_data, put_data;
unsigned char byte, *src_byte, *dest_byte;
src_byte = (unsigned char *)src;
dest_byte = (unsigned char *)dest;
dest_in_l1 = isram_check_addr(dest, n);
src_in_l1 = isram_check_addr(src, n);
need_data = true;
put_data = true;
for (count = 0; count < n; count++) {
if (src_in_l1) {
if (need_data) {
data_in = isram_read(src + count);
need_data = false;
}
if (ADDR2LAST(src + count))
need_data = true;
byte = (unsigned char)((data_in >> ADDR2OFFSET(src + count)) & 0xff);
} else {
/* src is in L2 or L3 - so just dereference*/
byte = src_byte[count];
}
if (dest_in_l1) {
if (put_data) {
data_out = isram_read(dest + count);
put_data = false;
}
data_out &= ~((uint64_t)0xff << ADDR2OFFSET(dest + count));
data_out |= ((uint64_t)byte << ADDR2OFFSET(dest + count));
if (ADDR2LAST(dest + count)) {
put_data = true;
isram_write(dest + count, data_out);
}
} else {
/* dest in L2 or L3 - so just dereference */
dest_byte[count] = byte;
}
}
/* make sure we dump the last byte if necessary */
if (dest_in_l1 && !put_data)
isram_write(dest + count, data_out);
return dest;
}
EXPORT_SYMBOL(isram_memcpy);
#ifdef CONFIG_BFIN_ISRAM_SELF_TEST
static int test_len = 0x20000;
static __init void hex_dump(unsigned char *buf, int len)
{
while (len--)
pr_cont("%02x", *buf++);
}
static __init int isram_read_test(char *sdram, void *l1inst)
{
int i, ret = 0;
uint64_t data1, data2;
pr_info("INFO: running isram_read tests\n");
/* setup some different data to play with */
for (i = 0; i < test_len; ++i)
sdram[i] = i % 255;
dma_memcpy(l1inst, sdram, test_len);
/* make sure we can read the L1 inst */
for (i = 0; i < test_len; i += sizeof(uint64_t)) {
data1 = isram_read(l1inst + i);
memcpy(&data2, sdram + i, sizeof(data2));
if (data1 != data2) {
pr_err("FAIL: isram_read(%p) returned %#llx but wanted %#llx\n",
l1inst + i, data1, data2);
++ret;
}
}
return ret;
}
static __init int isram_write_test(char *sdram, void *l1inst)
{
int i, ret = 0;
uint64_t data1, data2;
pr_info("INFO: running isram_write tests\n");
/* setup some different data to play with */
memset(sdram, 0, test_len * 2);
dma_memcpy(l1inst, sdram, test_len);
for (i = 0; i < test_len; ++i)
sdram[i] = i % 255;
/* make sure we can write the L1 inst */
for (i = 0; i < test_len; i += sizeof(uint64_t)) {
memcpy(&data1, sdram + i, sizeof(data1));
isram_write(l1inst + i, data1);
data2 = isram_read(l1inst + i);
if (data1 != data2) {
pr_err("FAIL: isram_write(%p, %#llx) != %#llx\n",
l1inst + i, data1, data2);
++ret;
}
}
dma_memcpy(sdram + test_len, l1inst, test_len);
if (memcmp(sdram, sdram + test_len, test_len)) {
pr_err("FAIL: isram_write() did not work properly\n");
++ret;
}
return ret;
}
static __init int
_isram_memcpy_test(char pattern, void *sdram, void *l1inst, const char *smemcpy,
void *(*fmemcpy)(void *, const void *, size_t))
{
memset(sdram, pattern, test_len);
fmemcpy(l1inst, sdram, test_len);
fmemcpy(sdram + test_len, l1inst, test_len);
if (memcmp(sdram, sdram + test_len, test_len)) {
pr_err("FAIL: %s(%p <=> %p, %#x) failed (data is %#x)\n",
smemcpy, l1inst, sdram, test_len, pattern);
return 1;
}
return 0;
}
#define _isram_memcpy_test(a, b, c, d) _isram_memcpy_test(a, b, c, #d, d)
static __init int isram_memcpy_test(char *sdram, void *l1inst)
{
int i, j, thisret, ret = 0;
/* check broad isram_memcpy() */
pr_info("INFO: running broad isram_memcpy tests\n");
for (i = 0xf; i >= 0; --i)
ret += _isram_memcpy_test(i, sdram, l1inst, isram_memcpy);
/* check read of small, unaligned, and hardware 64bit limits */
pr_info("INFO: running isram_memcpy (read) tests\n");
/* setup some different data to play with */
for (i = 0; i < test_len; ++i)
sdram[i] = i % 255;
dma_memcpy(l1inst, sdram, test_len);
thisret = 0;
for (i = 0; i < test_len - 32; ++i) {
unsigned char cmp[32];
for (j = 1; j <= 32; ++j) {
memset(cmp, 0, sizeof(cmp));
isram_memcpy(cmp, l1inst + i, j);
if (memcmp(cmp, sdram + i, j)) {
pr_err("FAIL: %p:", l1inst + 1);
hex_dump(cmp, j);
pr_cont(" SDRAM:");
hex_dump(sdram + i, j);
pr_cont("\n");
if (++thisret > 20) {
pr_err("FAIL: skipping remaining series\n");
i = test_len;
break;
}
}
}
}
ret += thisret;
/* check write of small, unaligned, and hardware 64bit limits */
pr_info("INFO: running isram_memcpy (write) tests\n");
memset(sdram + test_len, 0, test_len);
dma_memcpy(l1inst, sdram + test_len, test_len);
thisret = 0;
for (i = 0; i < test_len - 32; ++i) {
unsigned char cmp[32];
for (j = 1; j <= 32; ++j) {
isram_memcpy(l1inst + i, sdram + i, j);
dma_memcpy(cmp, l1inst + i, j);
if (memcmp(cmp, sdram + i, j)) {
pr_err("FAIL: %p:", l1inst + i);
hex_dump(cmp, j);
pr_cont(" SDRAM:");
hex_dump(sdram + i, j);
pr_cont("\n");
if (++thisret > 20) {
pr_err("FAIL: skipping remaining series\n");
i = test_len;
break;
}
}
}
}
ret += thisret;
return ret;
}
static __init int isram_test_init(void)
{
int ret;
char *sdram;
void *l1inst;
/* Try to test as much of L1SRAM as possible */
while (test_len) {
test_len >>= 1;
l1inst = l1_inst_sram_alloc(test_len);
if (l1inst)
break;
}
if (!l1inst) {
pr_warning("SKIP: could not allocate L1 inst\n");
return 0;
}
pr_info("INFO: testing %#x bytes (%p - %p)\n",
test_len, l1inst, l1inst + test_len);
sdram = kmalloc(test_len * 2, GFP_KERNEL);
if (!sdram) {
sram_free(l1inst);
pr_warning("SKIP: could not allocate sdram\n");
return 0;
}
/* sanity check initial L1 inst state */
ret = 1;
pr_info("INFO: running initial dma_memcpy checks %p\n", sdram);
if (_isram_memcpy_test(0xa, sdram, l1inst, dma_memcpy))
goto abort;
if (_isram_memcpy_test(0x5, sdram, l1inst, dma_memcpy))
goto abort;
ret = 0;
ret += isram_read_test(sdram, l1inst);
ret += isram_write_test(sdram, l1inst);
ret += isram_memcpy_test(sdram, l1inst);
abort:
sram_free(l1inst);
kfree(sdram);
if (ret)
return -EIO;
pr_info("PASS: all tests worked !\n");
return 0;
}
late_initcall(isram_test_init);
static __exit void isram_test_exit(void)
{
/* stub to allow unloading */
}
module_exit(isram_test_exit);
#endif