| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-Id: <1265225417.2116.2.camel@achroite.uk.solarflarecom.com>
Date: Wed, 03 Feb 2010 19:30:17 +0000
From: Ben Hutchings <bhutchings@...arflare.com>
To: David Miller <davem@...emloft.net>
Cc: netdev@...r.kernel.org, linux-net-drivers@...arflare.com
Subject: [PATCH 02/10] sfc: Handle firmware assertion failure while
resetting
From: Steve Hodgson <shodgson@...arflare.com>
This allows the driver to recover if the MC firmware has crashed due
to an assertion failure.
Signed-off-by: Ben Hutchings <bhutchings@...arflare.com>
---
drivers/net/sfc/mcdi.c | 63 ++++++++++++++++++++++++++++------------------
drivers/net/sfc/siena.c | 6 ++++
2 files changed, 44 insertions(+), 25 deletions(-)
diff --git a/drivers/net/sfc/mcdi.c b/drivers/net/sfc/mcdi.c
index 9f035b9..e9f0e5e 100644
--- a/drivers/net/sfc/mcdi.c
+++ b/drivers/net/sfc/mcdi.c
@@ -896,29 +896,27 @@ fail:
return rc;
}
-int efx_mcdi_handle_assertion(struct efx_nic *efx)
+static int efx_mcdi_read_assertion(struct efx_nic *efx)
{
- union {
- u8 asserts[MC_CMD_GET_ASSERTS_IN_LEN];
- u8 reboot[MC_CMD_REBOOT_IN_LEN];
- } inbuf;
- u8 assertion[MC_CMD_GET_ASSERTS_OUT_LEN];
+ u8 inbuf[MC_CMD_GET_ASSERTS_IN_LEN];
+ u8 outbuf[MC_CMD_GET_ASSERTS_OUT_LEN];
unsigned int flags, index, ofst;
const char *reason;
size_t outlen;
int retry;
int rc;
- /* Check if the MC is in the assertion handler, retrying twice. Once
+ /* Attempt to read any stored assertion state before we reboot
+ * the mcfw out of the assertion handler. Retry twice, once
* because a boot-time assertion might cause this command to fail
* with EINTR. And once again because GET_ASSERTS can race with
* MC_CMD_REBOOT running on the other port. */
retry = 2;
do {
- MCDI_SET_DWORD(inbuf.asserts, GET_ASSERTS_IN_CLEAR, 0);
+ MCDI_SET_DWORD(inbuf, GET_ASSERTS_IN_CLEAR, 1);
rc = efx_mcdi_rpc(efx, MC_CMD_GET_ASSERTS,
- inbuf.asserts, MC_CMD_GET_ASSERTS_IN_LEN,
- assertion, sizeof(assertion), &outlen);
+ inbuf, MC_CMD_GET_ASSERTS_IN_LEN,
+ outbuf, sizeof(outbuf), &outlen);
} while ((rc == -EINTR || rc == -EIO) && retry-- > 0);
if (rc)
@@ -926,21 +924,11 @@ int efx_mcdi_handle_assertion(struct efx_nic *efx)
if (outlen < MC_CMD_GET_ASSERTS_OUT_LEN)
return -EINVAL;
- flags = MCDI_DWORD(assertion, GET_ASSERTS_OUT_GLOBAL_FLAGS);
+ /* Print out any recorded assertion state */
+ flags = MCDI_DWORD(outbuf, GET_ASSERTS_OUT_GLOBAL_FLAGS);
if (flags == MC_CMD_GET_ASSERTS_FLAGS_NO_FAILS)
return 0;
- /* Reset the hardware atomically such that only one port with succeed.
- * This command will succeed if a reboot is no longer required (because
- * the other port did it first), but fail with EIO if it succeeds.
- */
- BUILD_BUG_ON(MC_CMD_REBOOT_OUT_LEN != 0);
- MCDI_SET_DWORD(inbuf.reboot, REBOOT_IN_FLAGS,
- MC_CMD_REBOOT_FLAGS_AFTER_ASSERTION);
- efx_mcdi_rpc(efx, MC_CMD_REBOOT, inbuf.reboot, MC_CMD_REBOOT_IN_LEN,
- NULL, 0, NULL);
-
- /* Print out the assertion */
reason = (flags == MC_CMD_GET_ASSERTS_FLAGS_SYS_FAIL)
? "system-level assertion"
: (flags == MC_CMD_GET_ASSERTS_FLAGS_THR_FAIL)
@@ -949,20 +937,45 @@ int efx_mcdi_handle_assertion(struct efx_nic *efx)
? "watchdog reset"
: "unknown assertion";
EFX_ERR(efx, "MCPU %s at PC = 0x%.8x in thread 0x%.8x\n", reason,
- MCDI_DWORD(assertion, GET_ASSERTS_OUT_SAVED_PC_OFFS),
- MCDI_DWORD(assertion, GET_ASSERTS_OUT_THREAD_OFFS));
+ MCDI_DWORD(outbuf, GET_ASSERTS_OUT_SAVED_PC_OFFS),
+ MCDI_DWORD(outbuf, GET_ASSERTS_OUT_THREAD_OFFS));
/* Print out the registers */
ofst = MC_CMD_GET_ASSERTS_OUT_GP_REGS_OFFS_OFST;
for (index = 1; index < 32; index++) {
EFX_ERR(efx, "R%.2d (?): 0x%.8x\n", index,
- MCDI_DWORD2(assertion, ofst));
+ MCDI_DWORD2(outbuf, ofst));
ofst += sizeof(efx_dword_t);
}
return 0;
}
+static void efx_mcdi_exit_assertion(struct efx_nic *efx)
+{
+ u8 inbuf[MC_CMD_REBOOT_IN_LEN];
+
+ /* Atomically reboot the mcfw out of the assertion handler */
+ BUILD_BUG_ON(MC_CMD_REBOOT_OUT_LEN != 0);
+ MCDI_SET_DWORD(inbuf, REBOOT_IN_FLAGS,
+ MC_CMD_REBOOT_FLAGS_AFTER_ASSERTION);
+ efx_mcdi_rpc(efx, MC_CMD_REBOOT, inbuf, MC_CMD_REBOOT_IN_LEN,
+ NULL, 0, NULL);
+}
+
+int efx_mcdi_handle_assertion(struct efx_nic *efx)
+{
+ int rc;
+
+ rc = efx_mcdi_read_assertion(efx);
+ if (rc)
+ return rc;
+
+ efx_mcdi_exit_assertion(efx);
+
+ return 0;
+}
+
void efx_mcdi_set_id_led(struct efx_nic *efx, enum efx_led_mode mode)
{
u8 inbuf[MC_CMD_SET_ID_LED_IN_LEN];
diff --git a/drivers/net/sfc/siena.c b/drivers/net/sfc/siena.c
index f8c6771..0e4c13a 100644
--- a/drivers/net/sfc/siena.c
+++ b/drivers/net/sfc/siena.c
@@ -181,6 +181,12 @@ static int siena_test_registers(struct efx_nic *efx)
static int siena_reset_hw(struct efx_nic *efx, enum reset_type method)
{
+ int rc;
+
+ /* Recover from a failed assertion pre-reset */
+ rc = efx_mcdi_handle_assertion(efx);
+ if (rc)
+ return rc;
if (method == RESET_TYPE_WORLD)
return efx_mcdi_reset_mc(efx);
--
1.6.2.5
--
Ben Hutchings, Senior Software Engineer, Solarflare Communications
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Powered by blists - more mailing lists