From: Shradha Shah <sshah@solarflare.com>
To: David Miller <davem@davemloft.net>
Cc: <netdev@vger.kernel.org>, <linux-net-drivers@solarflare.com>
Subject: [PATCH net-next 12/14] sfc: Cope with permissions enforcement added to firmware for SR-IOV
Date: Wed, 6 May 2015 00:59:18 +0100 [thread overview]
Message-ID: <55495956.1040808@solarflare.com> (raw)
In-Reply-To: <554957DF.2010307@solarflare.com>
From: Edward Cree <ecree@solarflare.com>
* Accept EPERM in some simple cases, the following cases are handled:
1) efx_mcdi_read_assertion()
Unprivileged PCI functions aren't allowed to GET_ASSERTS.
We return success as it's up to the primary PF to deal with asserts.
2) efx_mcdi_mon_probe() in efx_ef10_probe()
Unprivileged PCI functions aren't allowed to read sensor info, and
worrying about sensor data is the primary PF's job.
3) phy_op->reconfigure() in efx_init_port() and efx_reset_up()
Unprivileged functions aren't allowed to MC_CMD_SET_LINK, they just have
to accept the settings (including flow-control, which is what
efx_init_port() is worried about) they've been given.
4) Fallback to GET_WORKAROUNDS in efx_ef10_probe()
Unprivileged PCI functions aren't allowed to set workarounds. So if
efx_mcdi_set_workaround() fails EPERM, use efx_mcdi_get_workarounds()
to find out if workaround_35388 is enabled.
5) If DRV_ATTACH gets EPERM, try without specifying fw-variant
Unprivileged PCI functions have to use a FIRMWARE_ID of 0xffffffff
(MC_CMD_FW_DONT_CARE).
6) Don't try to exit_assertion unless one had fired
Previously we called efx_mcdi_exit_assertion even if
efx_mcdi_read_assertion had received MC_CMD_GET_ASSERTS_FLAGS_NO_FAILS.
This is unnecessary, and the resulting MC_CMD_REBOOT, even if the
AFTER_ASSERTION flag made it a no-op, would fail EPERM for unprivileged
PCI functions.
So make efx_mcdi_read_assertion return whether an assert happened, and only
call efx_mcdi_exit_assertion if it has.
Signed-off-by: Shradha Shah <sshah@solarflare.com>
---
drivers/net/ethernet/sfc/ef10.c | 16 ++++++-
drivers/net/ethernet/sfc/efx.c | 5 ++-
drivers/net/ethernet/sfc/mcdi.c | 82 ++++++++++++++++++++++++++++++------
drivers/net/ethernet/sfc/mcdi.h | 2 +
drivers/net/ethernet/sfc/mcdi_pcol.h | 23 ++++++++++
5 files changed, 111 insertions(+), 17 deletions(-)
diff --git a/drivers/net/ethernet/sfc/ef10.c b/drivers/net/ethernet/sfc/ef10.c
index 0dff812..ae0ed2a 100644
--- a/drivers/net/ethernet/sfc/ef10.c
+++ b/drivers/net/ethernet/sfc/ef10.c
@@ -286,10 +286,22 @@ static int efx_ef10_probe(struct efx_nic *efx)
goto fail3;
efx->timer_quantum_ns = 1536000 / rc; /* 1536 cycles */
- /* Check whether firmware supports bug 35388 workaround */
+ /* Check whether firmware supports bug 35388 workaround.
+ * First try to enable it, then if we get EPERM, just
+ * ask if it's already enabled
+ */
rc = efx_mcdi_set_workaround(efx, MC_CMD_WORKAROUND_BUG35388, true);
if (rc == 0)
nic_data->workaround_35388 = true;
+ else if (rc == -EPERM) {
+ unsigned int enabled;
+
+ rc = efx_mcdi_get_workarounds(efx, NULL, &enabled);
+ if (rc)
+ goto fail3;
+ nic_data->workaround_35388 = enabled &
+ MC_CMD_GET_WORKAROUNDS_OUT_BUG35388;
+ }
else if (rc != -ENOSYS && rc != -ENOENT)
goto fail3;
netif_dbg(efx, probe, efx->net_dev,
@@ -297,7 +309,7 @@ static int efx_ef10_probe(struct efx_nic *efx)
nic_data->workaround_35388 ? "en" : "dis");
rc = efx_mcdi_mon_probe(efx);
- if (rc)
+ if (rc && rc != -EPERM)
goto fail3;
efx_ptp_probe(efx, NULL);
diff --git a/drivers/net/ethernet/sfc/efx.c b/drivers/net/ethernet/sfc/efx.c
index 8b2edda..78f7760 100644
--- a/drivers/net/ethernet/sfc/efx.c
+++ b/drivers/net/ethernet/sfc/efx.c
@@ -1046,7 +1046,7 @@ static int efx_init_port(struct efx_nic *efx)
/* Ensure the PHY advertises the correct flow control settings */
rc = efx->phy_op->reconfigure(efx);
- if (rc)
+ if (rc && rc != -EPERM)
goto fail2;
mutex_unlock(&efx->mac_lock);
@@ -2429,7 +2429,8 @@ int efx_reset_up(struct efx_nic *efx, enum reset_type method, bool ok)
rc = efx->phy_op->init(efx);
if (rc)
goto fail;
- if (efx->phy_op->reconfigure(efx))
+ rc = efx->phy_op->reconfigure(efx);
+ if (rc && rc != -EPERM)
netif_err(efx, drv, efx->net_dev,
"could not restore PHY settings\n");
}
diff --git a/drivers/net/ethernet/sfc/mcdi.c b/drivers/net/ethernet/sfc/mcdi.c
index 6e0c789..1ffe289 100644
--- a/drivers/net/ethernet/sfc/mcdi.c
+++ b/drivers/net/ethernet/sfc/mcdi.c
@@ -1142,10 +1142,26 @@ static int efx_mcdi_drv_attach(struct efx_nic *efx, bool driver_operating,
MCDI_SET_DWORD(inbuf, DRV_ATTACH_IN_UPDATE, 1);
MCDI_SET_DWORD(inbuf, DRV_ATTACH_IN_FIRMWARE_ID, MC_CMD_FW_LOW_LATENCY);
- rc = efx_mcdi_rpc(efx, MC_CMD_DRV_ATTACH, inbuf, sizeof(inbuf),
- outbuf, sizeof(outbuf), &outlen);
- if (rc)
+ rc = efx_mcdi_rpc_quiet(efx, MC_CMD_DRV_ATTACH, inbuf, sizeof(inbuf),
+ outbuf, sizeof(outbuf), &outlen);
+ /* If we're not the primary PF, trying to ATTACH with a FIRMWARE_ID
+ * specified will fail with EPERM, and we have to tell the MC we don't
+ * care what firmware we get.
+ */
+ if (rc == -EPERM) {
+ netif_dbg(efx, probe, efx->net_dev,
+ "efx_mcdi_drv_attach with fw-variant setting failed EPERM, trying without it\n");
+ MCDI_SET_DWORD(inbuf, DRV_ATTACH_IN_FIRMWARE_ID,
+ MC_CMD_FW_DONT_CARE);
+ rc = efx_mcdi_rpc_quiet(efx, MC_CMD_DRV_ATTACH, inbuf,
+ sizeof(inbuf), outbuf, sizeof(outbuf),
+ &outlen);
+ }
+ if (rc) {
+ efx_mcdi_display_error(efx, MC_CMD_DRV_ATTACH, sizeof(inbuf),
+ outbuf, outlen, rc);
goto fail;
+ }
if (outlen < MC_CMD_DRV_ATTACH_OUT_LEN) {
rc = -EIO;
goto fail;
@@ -1377,6 +1393,9 @@ fail1:
return rc;
}
+/* Returns 1 if an assertion was read, 0 if no assertion had fired,
+ * negative on error.
+ */
static int efx_mcdi_read_assertion(struct efx_nic *efx)
{
MCDI_DECLARE_BUF(inbuf, MC_CMD_GET_ASSERTS_IN_LEN);
@@ -1398,6 +1417,8 @@ static int efx_mcdi_read_assertion(struct efx_nic *efx)
rc = efx_mcdi_rpc_quiet(efx, MC_CMD_GET_ASSERTS,
inbuf, MC_CMD_GET_ASSERTS_IN_LEN,
outbuf, sizeof(outbuf), &outlen);
+ if (rc == -EPERM)
+ return 0;
} while ((rc == -EINTR || rc == -EIO) && retry-- > 0);
if (rc) {
@@ -1435,24 +1456,31 @@ static int efx_mcdi_read_assertion(struct efx_nic *efx)
MCDI_ARRAY_DWORD(outbuf, GET_ASSERTS_OUT_GP_REGS_OFFS,
index));
- return 0;
+ return 1;
}
-static void efx_mcdi_exit_assertion(struct efx_nic *efx)
+static int efx_mcdi_exit_assertion(struct efx_nic *efx)
{
MCDI_DECLARE_BUF(inbuf, MC_CMD_REBOOT_IN_LEN);
+ int rc;
/* If the MC is running debug firmware, it might now be
* waiting for a debugger to attach, but we just want it to
* reboot. We set a flag that makes the command a no-op if it
- * has already done so. We don't know what return code to
- * expect (0 or -EIO), so ignore it.
+ * has already done so.
+ * The MCDI will thus return either 0 or -EIO.
*/
BUILD_BUG_ON(MC_CMD_REBOOT_OUT_LEN != 0);
MCDI_SET_DWORD(inbuf, REBOOT_IN_FLAGS,
MC_CMD_REBOOT_FLAGS_AFTER_ASSERTION);
- (void) efx_mcdi_rpc(efx, MC_CMD_REBOOT, inbuf, MC_CMD_REBOOT_IN_LEN,
- NULL, 0, NULL);
+ rc = efx_mcdi_rpc_quiet(efx, MC_CMD_REBOOT, inbuf, MC_CMD_REBOOT_IN_LEN,
+ NULL, 0, NULL);
+ if (rc == -EIO)
+ rc = 0;
+ if (rc)
+ efx_mcdi_display_error(efx, MC_CMD_REBOOT, MC_CMD_REBOOT_IN_LEN,
+ NULL, 0, rc);
+ return rc;
}
int efx_mcdi_handle_assertion(struct efx_nic *efx)
@@ -1460,12 +1488,10 @@ int efx_mcdi_handle_assertion(struct efx_nic *efx)
int rc;
rc = efx_mcdi_read_assertion(efx);
- if (rc)
+ if (rc <= 0)
return rc;
- efx_mcdi_exit_assertion(efx);
-
- return 0;
+ return efx_mcdi_exit_assertion(efx);
}
void efx_mcdi_set_id_led(struct efx_nic *efx, enum efx_led_mode mode)
@@ -1680,6 +1706,36 @@ int efx_mcdi_set_workaround(struct efx_nic *efx, u32 type, bool enabled)
NULL, 0, NULL);
}
+int efx_mcdi_get_workarounds(struct efx_nic *efx, unsigned int *impl_out,
+ unsigned int *enabled_out)
+{
+ MCDI_DECLARE_BUF_OUT_OR_ERR(outbuf, MC_CMD_GET_WORKAROUNDS_OUT_LEN);
+ size_t outlen;
+ int rc;
+
+ rc = efx_mcdi_rpc(efx, MC_CMD_GET_WORKAROUNDS, NULL, 0,
+ outbuf, sizeof(outbuf), &outlen);
+ if (rc)
+ goto fail;
+
+ if (outlen < MC_CMD_GET_WORKAROUNDS_OUT_LEN) {
+ rc = -EIO;
+ goto fail;
+ }
+
+ if (impl_out)
+ *impl_out = MCDI_DWORD(outbuf, GET_WORKAROUNDS_OUT_IMPLEMENTED);
+
+ if (enabled_out)
+ *enabled_out = MCDI_DWORD(outbuf, GET_WORKAROUNDS_OUT_ENABLED);
+
+ return 0;
+
+fail:
+ netif_err(efx, hw, efx->net_dev, "%s: failed rc=%d\n", __func__, rc);
+ return rc;
+}
+
#ifdef CONFIG_SFC_MTD
#define EFX_MCDI_NVRAM_LEN_MAX 128
diff --git a/drivers/net/ethernet/sfc/mcdi.h b/drivers/net/ethernet/sfc/mcdi.h
index 56465f7..5df1e98 100644
--- a/drivers/net/ethernet/sfc/mcdi.h
+++ b/drivers/net/ethernet/sfc/mcdi.h
@@ -339,6 +339,8 @@ bool efx_mcdi_mac_check_fault(struct efx_nic *efx);
enum reset_type efx_mcdi_map_reset_reason(enum reset_type reason);
int efx_mcdi_reset(struct efx_nic *efx, enum reset_type method);
int efx_mcdi_set_workaround(struct efx_nic *efx, u32 type, bool enabled);
+int efx_mcdi_get_workarounds(struct efx_nic *efx, unsigned int *impl_out,
+ unsigned int *enabled_out);
#ifdef CONFIG_SFC_MCDI_MON
int efx_mcdi_mon_probe(struct efx_nic *efx);
diff --git a/drivers/net/ethernet/sfc/mcdi_pcol.h b/drivers/net/ethernet/sfc/mcdi_pcol.h
index e028de1..4fa6eb2 100644
--- a/drivers/net/ethernet/sfc/mcdi_pcol.h
+++ b/drivers/net/ethernet/sfc/mcdi_pcol.h
@@ -1875,6 +1875,8 @@
#define MC_CMD_FW_FULL_FEATURED 0x0
/* enum: Prefer to use firmware with fewer features but lower latency */
#define MC_CMD_FW_LOW_LATENCY 0x1
+/* enum: Only this option is allowed for non-admin functions */
+#define MC_CMD_FW_DONT_CARE 0xffffffff
/* MC_CMD_DRV_ATTACH_OUT msgresponse */
#define MC_CMD_DRV_ATTACH_OUT_LEN 4
@@ -4087,6 +4089,27 @@
/***********************************/
+/* MC_CMD_GET_WORKAROUNDS
+ * Read the list of all implemented and all currently enabled workarounds. The
+ * enums here must correspond with those in MC_CMD_WORKAROUND.
+ */
+#define MC_CMD_GET_WORKAROUNDS 0x59
+
+/* MC_CMD_GET_WORKAROUNDS_OUT msgresponse */
+#define MC_CMD_GET_WORKAROUNDS_OUT_LEN 8
+/* Each workaround is represented by a single bit according to the enums below.
+ */
+#define MC_CMD_GET_WORKAROUNDS_OUT_IMPLEMENTED_OFST 0
+#define MC_CMD_GET_WORKAROUNDS_OUT_ENABLED_OFST 4
+/* enum: Bug 17230 work around. */
+#define MC_CMD_GET_WORKAROUNDS_OUT_BUG17230 0x2
+/* enum: Bug 35388 work around (unsafe EVQ writes). */
+#define MC_CMD_GET_WORKAROUNDS_OUT_BUG35388 0x4
+/* enum: Bug35017 workaround (A64 tables must be identity map) */
+#define MC_CMD_GET_WORKAROUNDS_OUT_BUG35017 0x8
+
+
+/***********************************/
/* MC_CMD_READ_REGS
* Get a dump of the MCPU registers
*/
next prev parent reply other threads:[~2015-05-05 23:59 UTC|newest]
Thread overview: 16+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-05-05 23:53 [PATCH net-next 00/14] sfc: Enabling EF10 Vf's, set up vswitching and bind the SFC driver to the VF's Shradha Shah
2015-05-05 23:55 ` [PATCH net-next 01/14] sfc: Own header for nic-specific sriov functions, single instance of netdev_ops and sriov removed from Falcon code Shradha Shah
2015-05-05 23:55 ` [PATCH net-next 02/14] sfc: Move and rename efx_vf struct to siena_vf Shradha Shah
2015-05-05 23:55 ` [PATCH net-next 03/14] sfc: Enable VF's via a write to the sysfs file sriov_numvfs Shradha Shah
2015-05-05 23:56 ` [PATCH net-next 04/14] sfc: Use MCDI to set FILTER_OP_IN_TX_DOMAIN Shradha Shah
2015-05-05 23:56 ` [PATCH net-next 05/14] sfc: Record [rt]x_dpcpu_fw_id in EF10 nic_data Shradha Shah
2015-05-05 23:57 ` [PATCH net-next 06/14] sfc: record the PF's vport ID in nic_data Shradha Shah
2015-05-05 23:57 ` [PATCH net-next 07/14] sfc: create VEB vswitch and vport above default firmware setup Shradha Shah
2015-05-05 23:57 ` [PATCH net-next 08/14] sfc: get the PF number and record in nic_data Shradha Shah
2015-05-05 23:58 ` [PATCH net-next 09/14] sfc: Prepare to bind the sfc driver to the VF Shradha Shah
2015-05-05 23:58 ` [PATCH net-next 10/14] sfc: create vports for VFs and assign random MAC addresses Shradha Shah
2015-05-05 23:58 ` [PATCH net-next 11/14] sfc: manually allocate and free vadaptors Shradha Shah
2015-05-05 23:59 ` Shradha Shah [this message]
2015-05-05 23:59 ` [PATCH net-next 13/14] sfc: Add use of shared RSS contexts Shradha Shah
2015-05-06 0:00 ` [PATCH net-next 14/14] sfc: Bind the sfc driver to any available VF's Shradha Shah
2015-05-09 20:18 ` [PATCH net-next 00/14] sfc: Enabling EF10 Vf's, set up vswitching and bind the SFC driver to the VF's David Miller
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=55495956.1040808@solarflare.com \
--to=sshah@solarflare.com \
--cc=davem@davemloft.net \
--cc=linux-net-drivers@solarflare.com \
--cc=netdev@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).