numam-dpdk/drivers/net/sfc/sfc_mcdi.c
Andrew Rybchenko e70c914539 net/sfc: avoid panic in the case of MCDI timeout
Implement dummy MCDI timeout handling which simply rejects
further MCDI requests.

Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Reviewed-by: Andy Moreton <amoreton@xilinx.com>
2020-09-30 19:19:08 +02:00

372 lines
9.3 KiB
C

/* SPDX-License-Identifier: BSD-3-Clause
*
* Copyright(c) 2019-2020 Xilinx, Inc.
* Copyright(c) 2016-2019 Solarflare Communications Inc.
*
* This software was jointly developed between OKTET Labs (under contract
* for Solarflare) and Solarflare Communications, Inc.
*/
#include <rte_cycles.h>
#include "efx.h"
#include "efx_mcdi.h"
#include "efx_regs_mcdi.h"
#include "sfc_mcdi.h"
#include "sfc.h"
#include "sfc_debug.h"
#include "sfc_log.h"
#include "sfc_ev.h"
#define SFC_EFX_MCDI_POLL_INTERVAL_MIN_US 10 /* 10us */
#define SFC_EFX_MCDI_POLL_INTERVAL_MAX_US (US_PER_S / 10) /* 100ms */
#define SFC_EFX_MCDI_WATCHDOG_INTERVAL_US (10 * US_PER_S) /* 10s */
#define sfc_efx_mcdi_log(mcdi, level, ...) \
do { \
const struct sfc_efx_mcdi *_mcdi = (mcdi); \
\
rte_log(level, _mcdi->logtype, \
RTE_FMT("%s" RTE_FMT_HEAD(__VA_ARGS__ ,) "\n", \
_mcdi->log_prefix, \
RTE_FMT_TAIL(__VA_ARGS__,))); \
} while (0)
#define sfc_efx_mcdi_crit(mcdi, ...) \
sfc_efx_mcdi_log(mcdi, RTE_LOG_CRIT, __VA_ARGS__)
#define sfc_efx_mcdi_err(mcdi, ...) \
sfc_efx_mcdi_log(mcdi, RTE_LOG_ERR, __VA_ARGS__)
#define sfc_efx_mcdi_warn(mcdi, ...) \
sfc_efx_mcdi_log(mcdi, RTE_LOG_WARNING, __VA_ARGS__)
#define sfc_efx_mcdi_info(mcdi, ...) \
sfc_efx_mcdi_log(mcdi, RTE_LOG_INFO, __VA_ARGS__)
/** Level value used by MCDI log statements */
#define SFC_EFX_LOG_LEVEL_MCDI RTE_LOG_INFO
#define sfc_efx_log_mcdi(mcdi, ...) \
sfc_efx_mcdi_log(mcdi, SFC_EFX_LOG_LEVEL_MCDI, __VA_ARGS__)
static void
sfc_efx_mcdi_timeout(struct sfc_adapter *sa)
{
struct sfc_efx_mcdi *mcdi = &sa->mcdi;
sfc_efx_mcdi_warn(mcdi, "MC TIMEOUT");
mcdi->state = SFC_EFX_MCDI_DEAD;
sfc_efx_mcdi_crit(mcdi,
"MCDI timeout handling is not implemented - NIC is unusable");
}
static inline boolean_t
sfc_efx_mcdi_proxy_event_available(struct sfc_adapter *sa)
{
struct sfc_efx_mcdi *mcdi = &sa->mcdi;
mcdi->proxy_handle = 0;
mcdi->proxy_result = ETIMEDOUT;
sfc_ev_mgmt_qpoll(sa);
if (mcdi->proxy_result != ETIMEDOUT)
return B_TRUE;
return B_FALSE;
}
static void
sfc_efx_mcdi_poll(struct sfc_adapter *sa, boolean_t proxy)
{
struct sfc_efx_mcdi *mcdi = &sa->mcdi;
efx_nic_t *enp;
unsigned int delay_total;
unsigned int delay_us;
boolean_t aborted __rte_unused;
delay_total = 0;
delay_us = SFC_EFX_MCDI_POLL_INTERVAL_MIN_US;
enp = mcdi->nic;
do {
boolean_t poll_completed;
poll_completed = (proxy) ? sfc_efx_mcdi_proxy_event_available(sa) :
efx_mcdi_request_poll(enp);
if (poll_completed)
return;
if (delay_total > SFC_EFX_MCDI_WATCHDOG_INTERVAL_US) {
if (!proxy) {
aborted = efx_mcdi_request_abort(enp);
SFC_ASSERT(aborted);
sfc_efx_mcdi_timeout(sa);
}
return;
}
rte_delay_us(delay_us);
delay_total += delay_us;
/* Exponentially back off the poll frequency */
RTE_BUILD_BUG_ON(SFC_EFX_MCDI_POLL_INTERVAL_MAX_US >
UINT_MAX / 2);
delay_us *= 2;
if (delay_us > SFC_EFX_MCDI_POLL_INTERVAL_MAX_US)
delay_us = SFC_EFX_MCDI_POLL_INTERVAL_MAX_US;
} while (1);
}
static void
sfc_efx_mcdi_execute(void *arg, efx_mcdi_req_t *emrp)
{
struct sfc_adapter *sa = (struct sfc_adapter *)arg;
struct sfc_efx_mcdi *mcdi = &sa->mcdi;
uint32_t proxy_handle;
if (mcdi->state == SFC_EFX_MCDI_DEAD) {
emrp->emr_rc = ENOEXEC;
return;
}
rte_spinlock_lock(&mcdi->lock);
SFC_ASSERT(mcdi->state == SFC_EFX_MCDI_INITIALIZED);
efx_mcdi_request_start(mcdi->nic, emrp, B_FALSE);
sfc_efx_mcdi_poll(sa, B_FALSE);
if (efx_mcdi_get_proxy_handle(mcdi->nic, emrp, &proxy_handle) == 0) {
/*
* Authorization is required for the MCDI request;
* wait for an MCDI proxy response event to bring
* a non-zero proxy handle (should be the same as
* the value obtained above) and operation status
*/
sfc_efx_mcdi_poll(sa, B_TRUE);
if ((mcdi->proxy_handle != 0) &&
(mcdi->proxy_handle != proxy_handle)) {
sfc_efx_mcdi_err(mcdi, "Unexpected MCDI proxy event");
emrp->emr_rc = EFAULT;
} else if (mcdi->proxy_result == 0) {
/*
* Authorization succeeded; re-issue the original
* request and poll for an ordinary MCDI response
*/
efx_mcdi_request_start(mcdi->nic, emrp, B_FALSE);
sfc_efx_mcdi_poll(sa, B_FALSE);
} else {
emrp->emr_rc = mcdi->proxy_result;
sfc_efx_mcdi_err(mcdi,
"MCDI proxy authorization failed (handle=%08x, result=%d)",
proxy_handle, mcdi->proxy_result);
}
}
rte_spinlock_unlock(&mcdi->lock);
}
static void
sfc_efx_mcdi_ev_cpl(void *arg)
{
struct sfc_adapter *sa = (struct sfc_adapter *)arg;
struct sfc_efx_mcdi *mcdi __rte_unused;
mcdi = &sa->mcdi;
SFC_ASSERT(mcdi->state == SFC_EFX_MCDI_INITIALIZED);
/* MCDI is polled, completions are not expected */
SFC_ASSERT(0);
}
static void
sfc_efx_mcdi_exception(void *arg, efx_mcdi_exception_t eme)
{
struct sfc_adapter *sa = (struct sfc_adapter *)arg;
struct sfc_efx_mcdi *mcdi = &sa->mcdi;
sfc_efx_mcdi_warn(mcdi, "MC %s",
(eme == EFX_MCDI_EXCEPTION_MC_REBOOT) ? "REBOOT" :
(eme == EFX_MCDI_EXCEPTION_MC_BADASSERT) ? "BADASSERT" : "UNKNOWN");
sfc_schedule_restart(sa);
}
#define SFC_MCDI_LOG_BUF_SIZE 128
static size_t
sfc_efx_mcdi_do_log(const struct sfc_efx_mcdi *mcdi,
char *buffer, void *data, size_t data_size,
size_t pfxsize, size_t position)
{
uint32_t *words = data;
/* Space separator plus 2 characters per byte */
const size_t word_str_space = 1 + 2 * sizeof(*words);
size_t i;
for (i = 0; i < data_size; i += sizeof(*words)) {
if (position + word_str_space >=
SFC_MCDI_LOG_BUF_SIZE) {
/* Flush at SFC_MCDI_LOG_BUF_SIZE with backslash
* at the end which is required by netlogdecode.
*/
buffer[position] = '\0';
sfc_efx_log_mcdi(mcdi, "%s \\", buffer);
/* Preserve prefix for the next log message */
position = pfxsize;
}
position += snprintf(buffer + position,
SFC_MCDI_LOG_BUF_SIZE - position,
" %08x", *words);
words++;
}
return position;
}
static void
sfc_efx_mcdi_logger(void *arg, efx_log_msg_t type,
void *header, size_t header_size,
void *data, size_t data_size)
{
struct sfc_adapter *sa = (struct sfc_adapter *)arg;
struct sfc_efx_mcdi *mcdi = &sa->mcdi;
char buffer[SFC_MCDI_LOG_BUF_SIZE];
size_t pfxsize;
size_t start;
/*
* Unlike the other cases, MCDI logging implies more onerous work
* needed to produce a message. If the dynamic log level prevents
* the end result from being printed, the CPU time will be wasted.
*
* To avoid wasting time, the actual level is examined in advance.
*/
if (rte_log_get_level(mcdi->logtype) < (int)SFC_EFX_LOG_LEVEL_MCDI)
return;
/* The format including prefix added by sfc_efx_log_mcdi() is the
* format consumed by the Solarflare netlogdecode tool.
*/
pfxsize = snprintf(buffer, sizeof(buffer), "MCDI RPC %s:",
type == EFX_LOG_MCDI_REQUEST ? "REQ" :
type == EFX_LOG_MCDI_RESPONSE ? "RESP" : "???");
start = sfc_efx_mcdi_do_log(mcdi, buffer, header, header_size,
pfxsize, pfxsize);
start = sfc_efx_mcdi_do_log(mcdi, buffer, data, data_size,
pfxsize, start);
if (start != pfxsize) {
buffer[start] = '\0';
sfc_efx_log_mcdi(mcdi, "%s", buffer);
}
}
static void
sfc_efx_mcdi_ev_proxy_response(void *arg, uint32_t handle, efx_rc_t result)
{
struct sfc_adapter *sa = (struct sfc_adapter *)arg;
struct sfc_efx_mcdi *mcdi = &sa->mcdi;
mcdi->proxy_handle = handle;
mcdi->proxy_result = result;
}
static int
sfc_efx_mcdi_init(struct sfc_adapter *sa, struct sfc_efx_mcdi *mcdi,
uint32_t logtype, const char *log_prefix, efx_nic_t *nic)
{
size_t max_msg_size;
efx_mcdi_transport_t *emtp;
int rc;
SFC_ASSERT(mcdi->state == SFC_EFX_MCDI_UNINITIALIZED);
rte_spinlock_init(&mcdi->lock);
mcdi->nic = nic;
mcdi->state = SFC_EFX_MCDI_INITIALIZED;
mcdi->logtype = logtype;
mcdi->log_prefix = log_prefix;
max_msg_size = sizeof(uint32_t) + MCDI_CTL_SDU_LEN_MAX_V2;
rc = sfc_dma_alloc(sa, "mcdi", 0, max_msg_size, sa->socket_id,
&mcdi->mem);
if (rc != 0)
goto fail_dma_alloc;
emtp = &mcdi->transport;
emtp->emt_context = sa;
emtp->emt_dma_mem = &mcdi->mem;
emtp->emt_execute = sfc_efx_mcdi_execute;
emtp->emt_ev_cpl = sfc_efx_mcdi_ev_cpl;
emtp->emt_exception = sfc_efx_mcdi_exception;
emtp->emt_logger = sfc_efx_mcdi_logger;
emtp->emt_ev_proxy_response = sfc_efx_mcdi_ev_proxy_response;
sfc_efx_mcdi_info(mcdi, "init MCDI");
rc = efx_mcdi_init(mcdi->nic, emtp);
if (rc != 0)
goto fail_mcdi_init;
return 0;
fail_mcdi_init:
memset(emtp, 0, sizeof(*emtp));
sfc_dma_free(sa, &mcdi->mem);
fail_dma_alloc:
mcdi->state = SFC_EFX_MCDI_UNINITIALIZED;
return rc;
}
static void
sfc_efx_mcdi_fini(struct sfc_adapter *sa, struct sfc_efx_mcdi *mcdi)
{
efx_mcdi_transport_t *emtp;
emtp = &mcdi->transport;
rte_spinlock_lock(&mcdi->lock);
SFC_ASSERT(mcdi->state == SFC_EFX_MCDI_INITIALIZED ||
mcdi->state == SFC_EFX_MCDI_DEAD);
mcdi->state = SFC_EFX_MCDI_UNINITIALIZED;
sfc_efx_mcdi_info(mcdi, "fini MCDI");
efx_mcdi_fini(mcdi->nic);
memset(emtp, 0, sizeof(*emtp));
rte_spinlock_unlock(&mcdi->lock);
sfc_dma_free(sa, &mcdi->mem);
}
int
sfc_mcdi_init(struct sfc_adapter *sa)
{
uint32_t logtype;
sfc_log_init(sa, "entry");
logtype = sfc_register_logtype(&sa->priv.shared->pci_addr,
SFC_LOGTYPE_MCDI_STR,
RTE_LOG_NOTICE);
return sfc_efx_mcdi_init(sa, &sa->mcdi, logtype,
sa->priv.shared->log_prefix, sa->nic);
}
void
sfc_mcdi_fini(struct sfc_adapter *sa)
{
sfc_log_init(sa, "entry");
sfc_efx_mcdi_fini(sa, &sa->mcdi);
}