u-boot/drivers/crypto/fsl/jr.c
Saksham Jain 8a6f83dcb8 crypto/fsl: Make CAAM transactions cacheable
This commit solves CAAM coherency issue on ls2080. When caches are
enabled and CAAM's DMA's AXI transcations are not made cacheable,
Core reads/writes data from/to caches and CAAM does from main memory.
This forces data flushes to synchronize various data structures. But
even if any data in proximity of these structures is read by core,
these structures again are fetched in caches.

To avoid this problem, either all the data that CAAM accesses can be
made cache line aligned or CAAM transcations can be made cacheable.

So, this commit makes CAAM transcations as write back with write and
read allocate.

Signed-off-by: Saksham Jain <saksham.jain@nxp.com>
Reviewed-by: York Sun <york.sun@nxp.com>
2016-03-29 08:46:22 -07:00

594 lines
14 KiB
C

/*
* Copyright 2008-2014 Freescale Semiconductor, Inc.
*
* SPDX-License-Identifier: GPL-2.0+
*
* Based on CAAM driver in drivers/crypto/caam in Linux
*/
#include <common.h>
#include <malloc.h>
#include "fsl_sec.h"
#include "jr.h"
#include "jobdesc.h"
#include "desc_constr.h"
#ifdef CONFIG_FSL_CORENET
#include <asm/fsl_pamu.h>
#endif
#define CIRC_CNT(head, tail, size) (((head) - (tail)) & (size - 1))
#define CIRC_SPACE(head, tail, size) CIRC_CNT((tail), (head) + 1, (size))
struct jobring jr;
static inline void start_jr0(void)
{
ccsr_sec_t *sec = (void *)CONFIG_SYS_FSL_SEC_ADDR;
u32 ctpr_ms = sec_in32(&sec->ctpr_ms);
u32 scfgr = sec_in32(&sec->scfgr);
if (ctpr_ms & SEC_CTPR_MS_VIRT_EN_INCL) {
/* VIRT_EN_INCL = 1 & VIRT_EN_POR = 1 or
* VIRT_EN_INCL = 1 & VIRT_EN_POR = 0 & SEC_SCFGR_VIRT_EN = 1
*/
if ((ctpr_ms & SEC_CTPR_MS_VIRT_EN_POR) ||
(!(ctpr_ms & SEC_CTPR_MS_VIRT_EN_POR) &&
(scfgr & SEC_SCFGR_VIRT_EN)))
sec_out32(&sec->jrstartr, CONFIG_JRSTARTR_JR0);
} else {
/* VIRT_EN_INCL = 0 && VIRT_EN_POR_VALUE = 1 */
if (ctpr_ms & SEC_CTPR_MS_VIRT_EN_POR)
sec_out32(&sec->jrstartr, CONFIG_JRSTARTR_JR0);
}
}
static inline void jr_reset_liodn(void)
{
ccsr_sec_t *sec = (void *)CONFIG_SYS_FSL_SEC_ADDR;
sec_out32(&sec->jrliodnr[0].ls, 0);
}
static inline void jr_disable_irq(void)
{
struct jr_regs *regs = (struct jr_regs *)CONFIG_SYS_FSL_JR0_ADDR;
uint32_t jrcfg = sec_in32(&regs->jrcfg1);
jrcfg = jrcfg | JR_INTMASK;
sec_out32(&regs->jrcfg1, jrcfg);
}
static void jr_initregs(void)
{
struct jr_regs *regs = (struct jr_regs *)CONFIG_SYS_FSL_JR0_ADDR;
phys_addr_t ip_base = virt_to_phys((void *)jr.input_ring);
phys_addr_t op_base = virt_to_phys((void *)jr.output_ring);
#ifdef CONFIG_PHYS_64BIT
sec_out32(&regs->irba_h, ip_base >> 32);
#else
sec_out32(&regs->irba_h, 0x0);
#endif
sec_out32(&regs->irba_l, (uint32_t)ip_base);
#ifdef CONFIG_PHYS_64BIT
sec_out32(&regs->orba_h, op_base >> 32);
#else
sec_out32(&regs->orba_h, 0x0);
#endif
sec_out32(&regs->orba_l, (uint32_t)op_base);
sec_out32(&regs->ors, JR_SIZE);
sec_out32(&regs->irs, JR_SIZE);
if (!jr.irq)
jr_disable_irq();
}
static int jr_init(void)
{
memset(&jr, 0, sizeof(struct jobring));
jr.jq_id = DEFAULT_JR_ID;
jr.irq = DEFAULT_IRQ;
#ifdef CONFIG_FSL_CORENET
jr.liodn = DEFAULT_JR_LIODN;
#endif
jr.size = JR_SIZE;
jr.input_ring = (dma_addr_t *)memalign(ARCH_DMA_MINALIGN,
JR_SIZE * sizeof(dma_addr_t));
if (!jr.input_ring)
return -1;
jr.op_size = roundup(JR_SIZE * sizeof(struct op_ring),
ARCH_DMA_MINALIGN);
jr.output_ring =
(struct op_ring *)memalign(ARCH_DMA_MINALIGN, jr.op_size);
if (!jr.output_ring)
return -1;
memset(jr.input_ring, 0, JR_SIZE * sizeof(dma_addr_t));
memset(jr.output_ring, 0, jr.op_size);
start_jr0();
jr_initregs();
return 0;
}
static int jr_sw_cleanup(void)
{
jr.head = 0;
jr.tail = 0;
jr.read_idx = 0;
jr.write_idx = 0;
memset(jr.info, 0, sizeof(jr.info));
memset(jr.input_ring, 0, jr.size * sizeof(dma_addr_t));
memset(jr.output_ring, 0, jr.size * sizeof(struct op_ring));
return 0;
}
static int jr_hw_reset(void)
{
struct jr_regs *regs = (struct jr_regs *)CONFIG_SYS_FSL_JR0_ADDR;
uint32_t timeout = 100000;
uint32_t jrint, jrcr;
sec_out32(&regs->jrcr, JRCR_RESET);
do {
jrint = sec_in32(&regs->jrint);
} while (((jrint & JRINT_ERR_HALT_MASK) ==
JRINT_ERR_HALT_INPROGRESS) && --timeout);
jrint = sec_in32(&regs->jrint);
if (((jrint & JRINT_ERR_HALT_MASK) !=
JRINT_ERR_HALT_INPROGRESS) && timeout == 0)
return -1;
timeout = 100000;
sec_out32(&regs->jrcr, JRCR_RESET);
do {
jrcr = sec_in32(&regs->jrcr);
} while ((jrcr & JRCR_RESET) && --timeout);
if (timeout == 0)
return -1;
return 0;
}
/* -1 --- error, can't enqueue -- no space available */
static int jr_enqueue(uint32_t *desc_addr,
void (*callback)(uint32_t status, void *arg),
void *arg)
{
struct jr_regs *regs = (struct jr_regs *)CONFIG_SYS_FSL_JR0_ADDR;
int head = jr.head;
uint32_t desc_word;
int length = desc_len(desc_addr);
int i;
#ifdef CONFIG_PHYS_64BIT
uint32_t *addr_hi, *addr_lo;
#endif
/* The descriptor must be submitted to SEC block as per endianness
* of the SEC Block.
* So, if the endianness of Core and SEC block is different, each word
* of the descriptor will be byte-swapped.
*/
for (i = 0; i < length; i++) {
desc_word = desc_addr[i];
sec_out32((uint32_t *)&desc_addr[i], desc_word);
}
phys_addr_t desc_phys_addr = virt_to_phys(desc_addr);
if (sec_in32(&regs->irsa) == 0 ||
CIRC_SPACE(jr.head, jr.tail, jr.size) <= 0)
return -1;
jr.info[head].desc_phys_addr = desc_phys_addr;
jr.info[head].callback = (void *)callback;
jr.info[head].arg = arg;
jr.info[head].op_done = 0;
unsigned long start = (unsigned long)&jr.info[head] &
~(ARCH_DMA_MINALIGN - 1);
unsigned long end = ALIGN((unsigned long)&jr.info[head] +
sizeof(struct jr_info), ARCH_DMA_MINALIGN);
flush_dcache_range(start, end);
#ifdef CONFIG_PHYS_64BIT
/* Write the 64 bit Descriptor address on Input Ring.
* The 32 bit hign and low part of the address will
* depend on endianness of SEC block.
*/
#ifdef CONFIG_SYS_FSL_SEC_LE
addr_lo = (uint32_t *)(&jr.input_ring[head]);
addr_hi = (uint32_t *)(&jr.input_ring[head]) + 1;
#elif defined(CONFIG_SYS_FSL_SEC_BE)
addr_hi = (uint32_t *)(&jr.input_ring[head]);
addr_lo = (uint32_t *)(&jr.input_ring[head]) + 1;
#endif /* ifdef CONFIG_SYS_FSL_SEC_LE */
sec_out32(addr_hi, (uint32_t)(desc_phys_addr >> 32));
sec_out32(addr_lo, (uint32_t)(desc_phys_addr));
#else
/* Write the 32 bit Descriptor address on Input Ring. */
sec_out32(&jr.input_ring[head], desc_phys_addr);
#endif /* ifdef CONFIG_PHYS_64BIT */
start = (unsigned long)&jr.input_ring[head] & ~(ARCH_DMA_MINALIGN - 1);
end = ALIGN((unsigned long)&jr.input_ring[head] +
sizeof(dma_addr_t), ARCH_DMA_MINALIGN);
flush_dcache_range(start, end);
jr.head = (head + 1) & (jr.size - 1);
/* Invalidate output ring */
start = (unsigned long)jr.output_ring &
~(ARCH_DMA_MINALIGN - 1);
end = ALIGN((unsigned long)jr.output_ring + jr.op_size,
ARCH_DMA_MINALIGN);
invalidate_dcache_range(start, end);
sec_out32(&regs->irja, 1);
return 0;
}
static int jr_dequeue(void)
{
struct jr_regs *regs = (struct jr_regs *)CONFIG_SYS_FSL_JR0_ADDR;
int head = jr.head;
int tail = jr.tail;
int idx, i, found;
void (*callback)(uint32_t status, void *arg);
void *arg = NULL;
#ifdef CONFIG_PHYS_64BIT
uint32_t *addr_hi, *addr_lo;
#else
uint32_t *addr;
#endif
while (sec_in32(&regs->orsf) && CIRC_CNT(jr.head, jr.tail, jr.size)) {
found = 0;
phys_addr_t op_desc;
#ifdef CONFIG_PHYS_64BIT
/* Read the 64 bit Descriptor address from Output Ring.
* The 32 bit hign and low part of the address will
* depend on endianness of SEC block.
*/
#ifdef CONFIG_SYS_FSL_SEC_LE
addr_lo = (uint32_t *)(&jr.output_ring[jr.tail].desc);
addr_hi = (uint32_t *)(&jr.output_ring[jr.tail].desc) + 1;
#elif defined(CONFIG_SYS_FSL_SEC_BE)
addr_hi = (uint32_t *)(&jr.output_ring[jr.tail].desc);
addr_lo = (uint32_t *)(&jr.output_ring[jr.tail].desc) + 1;
#endif /* ifdef CONFIG_SYS_FSL_SEC_LE */
op_desc = ((u64)sec_in32(addr_hi) << 32) |
((u64)sec_in32(addr_lo));
#else
/* Read the 32 bit Descriptor address from Output Ring. */
addr = (uint32_t *)&jr.output_ring[jr.tail].desc;
op_desc = sec_in32(addr);
#endif /* ifdef CONFIG_PHYS_64BIT */
uint32_t status = sec_in32(&jr.output_ring[jr.tail].status);
for (i = 0; CIRC_CNT(head, tail + i, jr.size) >= 1; i++) {
idx = (tail + i) & (jr.size - 1);
if (op_desc == jr.info[idx].desc_phys_addr) {
found = 1;
break;
}
}
/* Error condition if match not found */
if (!found)
return -1;
jr.info[idx].op_done = 1;
callback = (void *)jr.info[idx].callback;
arg = jr.info[idx].arg;
/* When the job on tail idx gets done, increment
* tail till the point where job completed out of oredr has
* been taken into account
*/
if (idx == tail)
do {
tail = (tail + 1) & (jr.size - 1);
} while (jr.info[tail].op_done);
jr.tail = tail;
jr.read_idx = (jr.read_idx + 1) & (jr.size - 1);
sec_out32(&regs->orjr, 1);
jr.info[idx].op_done = 0;
callback(status, arg);
}
return 0;
}
static void desc_done(uint32_t status, void *arg)
{
struct result *x = arg;
x->status = status;
caam_jr_strstatus(status);
x->done = 1;
}
int run_descriptor_jr(uint32_t *desc)
{
unsigned long long timeval = get_ticks();
unsigned long long timeout = usec2ticks(CONFIG_SEC_DEQ_TIMEOUT);
struct result op;
int ret = 0;
memset(&op, 0, sizeof(op));
ret = jr_enqueue(desc, desc_done, &op);
if (ret) {
debug("Error in SEC enq\n");
ret = JQ_ENQ_ERR;
goto out;
}
timeval = get_ticks();
timeout = usec2ticks(CONFIG_SEC_DEQ_TIMEOUT);
while (op.done != 1) {
ret = jr_dequeue();
if (ret) {
debug("Error in SEC deq\n");
ret = JQ_DEQ_ERR;
goto out;
}
if ((get_ticks() - timeval) > timeout) {
debug("SEC Dequeue timed out\n");
ret = JQ_DEQ_TO_ERR;
goto out;
}
}
if (op.status) {
debug("Error %x\n", op.status);
ret = op.status;
}
out:
return ret;
}
int jr_reset(void)
{
if (jr_hw_reset() < 0)
return -1;
/* Clean up the jobring structure maintained by software */
jr_sw_cleanup();
return 0;
}
int sec_reset(void)
{
ccsr_sec_t *sec = (void *)CONFIG_SYS_FSL_SEC_ADDR;
uint32_t mcfgr = sec_in32(&sec->mcfgr);
uint32_t timeout = 100000;
mcfgr |= MCFGR_SWRST;
sec_out32(&sec->mcfgr, mcfgr);
mcfgr |= MCFGR_DMA_RST;
sec_out32(&sec->mcfgr, mcfgr);
do {
mcfgr = sec_in32(&sec->mcfgr);
} while ((mcfgr & MCFGR_DMA_RST) == MCFGR_DMA_RST && --timeout);
if (timeout == 0)
return -1;
timeout = 100000;
do {
mcfgr = sec_in32(&sec->mcfgr);
} while ((mcfgr & MCFGR_SWRST) == MCFGR_SWRST && --timeout);
if (timeout == 0)
return -1;
return 0;
}
static int instantiate_rng(void)
{
struct result op;
u32 *desc;
u32 rdsta_val;
int ret = 0;
ccsr_sec_t __iomem *sec =
(ccsr_sec_t __iomem *)CONFIG_SYS_FSL_SEC_ADDR;
struct rng4tst __iomem *rng =
(struct rng4tst __iomem *)&sec->rng;
memset(&op, 0, sizeof(struct result));
desc = memalign(ARCH_DMA_MINALIGN, sizeof(uint32_t) * 6);
if (!desc) {
printf("cannot allocate RNG init descriptor memory\n");
return -1;
}
inline_cnstr_jobdesc_rng_instantiation(desc);
int size = roundup(sizeof(uint32_t) * 6, ARCH_DMA_MINALIGN);
flush_dcache_range((unsigned long)desc,
(unsigned long)desc + size);
ret = run_descriptor_jr(desc);
if (ret)
printf("RNG: Instantiation failed with error %x\n", ret);
rdsta_val = sec_in32(&rng->rdsta);
if (op.status || !(rdsta_val & RNG_STATE0_HANDLE_INSTANTIATED))
return -1;
return ret;
}
static u8 get_rng_vid(void)
{
ccsr_sec_t *sec = (void *)CONFIG_SYS_FSL_SEC_ADDR;
u32 cha_vid = sec_in32(&sec->chavid_ls);
return (cha_vid & SEC_CHAVID_RNG_LS_MASK) >> SEC_CHAVID_LS_RNG_SHIFT;
}
/*
* By default, the TRNG runs for 200 clocks per sample;
* 1200 clocks per sample generates better entropy.
*/
static void kick_trng(int ent_delay)
{
ccsr_sec_t __iomem *sec =
(ccsr_sec_t __iomem *)CONFIG_SYS_FSL_SEC_ADDR;
struct rng4tst __iomem *rng =
(struct rng4tst __iomem *)&sec->rng;
u32 val;
/* put RNG4 into program mode */
sec_setbits32(&rng->rtmctl, RTMCTL_PRGM);
/* rtsdctl bits 0-15 contain "Entropy Delay, which defines the
* length (in system clocks) of each Entropy sample taken
* */
val = sec_in32(&rng->rtsdctl);
val = (val & ~RTSDCTL_ENT_DLY_MASK) |
(ent_delay << RTSDCTL_ENT_DLY_SHIFT);
sec_out32(&rng->rtsdctl, val);
/* min. freq. count, equal to 1/4 of the entropy sample length */
sec_out32(&rng->rtfreqmin, ent_delay >> 2);
/* disable maximum frequency count */
sec_out32(&rng->rtfreqmax, RTFRQMAX_DISABLE);
/*
* select raw sampling in both entropy shifter
* and statistical checker
*/
sec_setbits32(&rng->rtmctl, RTMCTL_SAMP_MODE_RAW_ES_SC);
/* put RNG4 into run mode */
sec_clrbits32(&rng->rtmctl, RTMCTL_PRGM);
}
static int rng_init(void)
{
int ret, ent_delay = RTSDCTL_ENT_DLY_MIN;
ccsr_sec_t __iomem *sec =
(ccsr_sec_t __iomem *)CONFIG_SYS_FSL_SEC_ADDR;
struct rng4tst __iomem *rng =
(struct rng4tst __iomem *)&sec->rng;
u32 rdsta = sec_in32(&rng->rdsta);
/* Check if RNG state 0 handler is already instantiated */
if (rdsta & RNG_STATE0_HANDLE_INSTANTIATED)
return 0;
do {
/*
* If either of the SH's were instantiated by somebody else
* then it is assumed that the entropy
* parameters are properly set and thus the function
* setting these (kick_trng(...)) is skipped.
* Also, if a handle was instantiated, do not change
* the TRNG parameters.
*/
kick_trng(ent_delay);
ent_delay += 400;
/*
* if instantiate_rng(...) fails, the loop will rerun
* and the kick_trng(...) function will modfiy the
* upper and lower limits of the entropy sampling
* interval, leading to a sucessful initialization of
* the RNG.
*/
ret = instantiate_rng();
} while ((ret == -1) && (ent_delay < RTSDCTL_ENT_DLY_MAX));
if (ret) {
printf("RNG: Failed to instantiate RNG\n");
return ret;
}
/* Enable RDB bit so that RNG works faster */
sec_setbits32(&sec->scfgr, SEC_SCFGR_RDBENABLE);
return ret;
}
int sec_init(void)
{
ccsr_sec_t *sec = (void *)CONFIG_SYS_FSL_SEC_ADDR;
uint32_t mcr = sec_in32(&sec->mcfgr);
int ret = 0;
#ifdef CONFIG_FSL_CORENET
uint32_t liodnr;
uint32_t liodn_ns;
uint32_t liodn_s;
#endif
/*
* Modifying CAAM Read/Write Attributes
* For LS2080A and LS2085A
* For AXI Write - Cacheable, Write Back, Write allocate
* For AXI Read - Cacheable, Read allocate
* Only For LS2080a and LS2085a, to solve CAAM coherency issues
*/
#if defined(CONFIG_LS2080A) || defined(CONFIG_LS2085A)
mcr = (mcr & ~MCFGR_AWCACHE_MASK) | (0xb << MCFGR_AWCACHE_SHIFT);
mcr = (mcr & ~MCFGR_ARCACHE_MASK) | (0x6 << MCFGR_ARCACHE_SHIFT);
#else
mcr = (mcr & ~MCFGR_AWCACHE_MASK) | (0x2 << MCFGR_AWCACHE_SHIFT);
#endif
#ifdef CONFIG_PHYS_64BIT
mcr |= (1 << MCFGR_PS_SHIFT);
#endif
sec_out32(&sec->mcfgr, mcr);
#ifdef CONFIG_FSL_CORENET
liodnr = sec_in32(&sec->jrliodnr[0].ls);
liodn_ns = (liodnr & JRNSLIODN_MASK) >> JRNSLIODN_SHIFT;
liodn_s = (liodnr & JRSLIODN_MASK) >> JRSLIODN_SHIFT;
#endif
ret = jr_init();
if (ret < 0) {
printf("SEC initialization failed\n");
return -1;
}
#ifdef CONFIG_FSL_CORENET
ret = sec_config_pamu_table(liodn_ns, liodn_s);
if (ret < 0)
return -1;
pamu_enable();
#endif
if (get_rng_vid() >= 4) {
if (rng_init() < 0) {
printf("RNG instantiation failed\n");
return -1;
}
printf("SEC: RNG instantiated\n");
}
return ret;
}