* [PATCH v4 01/28] e1000e: Fix the code style
2023-01-30 13:46 [PATCH v4 00/28] e1000x cleanups (preliminary for IGB) Akihiko Odaki
@ 2023-01-30 13:46 ` Akihiko Odaki
2023-01-30 13:46 ` [PATCH v4 02/28] hw/net: Add more MII definitions Akihiko Odaki
` (26 subsequent siblings)
27 siblings, 0 replies; 33+ messages in thread
From: Akihiko Odaki @ 2023-01-30 13:46 UTC (permalink / raw)
Cc: Jason Wang, Dmitry Fleytman, Thomas Huth, Laurent Vivier,
Paolo Bonzini, qemu-devel, qemu-ppc, Michael S . Tsirkin,
Yan Vugenfirer, Yuri Benditovich, Sriram Yagnaraman,
Alexander Bulekov, Akihiko Odaki, Philippe Mathieu-Daudé
igb implementation first starts off by copying e1000e code. Correct the
code style before that.
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
---
hw/net/e1000.c | 41 ++++++++--------
hw/net/e1000e.c | 72 ++++++++++++++--------------
hw/net/e1000e_core.c | 103 ++++++++++++++++++++++-------------------
hw/net/e1000e_core.h | 66 +++++++++++++-------------
hw/net/e1000x_common.h | 44 +++++++++---------
5 files changed, 168 insertions(+), 158 deletions(-)
diff --git a/hw/net/e1000.c b/hw/net/e1000.c
index 7efb8a4c52..8ee30aa37c 100644
--- a/hw/net/e1000.c
+++ b/hw/net/e1000.c
@@ -808,10 +808,11 @@ receive_filter(E1000State *s, const uint8_t *buf, int size)
if (e1000x_is_vlan_packet(buf, le16_to_cpu(s->mac_reg[VET])) &&
e1000x_vlan_rx_filter_enabled(s->mac_reg)) {
uint16_t vid = lduw_be_p(buf + 14);
- uint32_t vfta = ldl_le_p((uint32_t*)(s->mac_reg + VFTA) +
+ uint32_t vfta = ldl_le_p((uint32_t *)(s->mac_reg + VFTA) +
((vid >> 5) & 0x7f));
- if ((vfta & (1 << (vid & 0x1f))) == 0)
+ if ((vfta & (1 << (vid & 0x1f))) == 0) {
return 0;
+ }
}
if (!isbcast && !ismcast && (rctl & E1000_RCTL_UPE)) { /* promiscuous ucast */
@@ -1220,16 +1221,16 @@ static const readops macreg_readops[] = {
[TDFPC] = mac_low13_read,
[AIT] = mac_low16_read,
- [CRCERRS ... MPC] = &mac_readreg,
- [IP6AT ... IP6AT+3] = &mac_readreg, [IP4AT ... IP4AT+6] = &mac_readreg,
- [FFLT ... FFLT+6] = &mac_low11_read,
- [RA ... RA+31] = &mac_readreg,
- [WUPM ... WUPM+31] = &mac_readreg,
- [MTA ... MTA+127] = &mac_readreg,
- [VFTA ... VFTA+127] = &mac_readreg,
- [FFMT ... FFMT+254] = &mac_low4_read,
- [FFVT ... FFVT+254] = &mac_readreg,
- [PBM ... PBM+16383] = &mac_readreg,
+ [CRCERRS ... MPC] = &mac_readreg,
+ [IP6AT ... IP6AT + 3] = &mac_readreg, [IP4AT ... IP4AT + 6] = &mac_readreg,
+ [FFLT ... FFLT + 6] = &mac_low11_read,
+ [RA ... RA + 31] = &mac_readreg,
+ [WUPM ... WUPM + 31] = &mac_readreg,
+ [MTA ... MTA + 127] = &mac_readreg,
+ [VFTA ... VFTA + 127] = &mac_readreg,
+ [FFMT ... FFMT + 254] = &mac_low4_read,
+ [FFVT ... FFVT + 254] = &mac_readreg,
+ [PBM ... PBM + 16383] = &mac_readreg,
};
enum { NREADOPS = ARRAY_SIZE(macreg_readops) };
@@ -1252,14 +1253,14 @@ static const writeops macreg_writeops[] = {
[RDTR] = set_16bit, [RADV] = set_16bit, [TADV] = set_16bit,
[ITR] = set_16bit,
- [IP6AT ... IP6AT+3] = &mac_writereg, [IP4AT ... IP4AT+6] = &mac_writereg,
- [FFLT ... FFLT+6] = &mac_writereg,
- [RA ... RA+31] = &mac_writereg,
- [WUPM ... WUPM+31] = &mac_writereg,
- [MTA ... MTA+127] = &mac_writereg,
- [VFTA ... VFTA+127] = &mac_writereg,
- [FFMT ... FFMT+254] = &mac_writereg, [FFVT ... FFVT+254] = &mac_writereg,
- [PBM ... PBM+16383] = &mac_writereg,
+ [IP6AT ... IP6AT + 3] = &mac_writereg, [IP4AT ... IP4AT + 6] = &mac_writereg,
+ [FFLT ... FFLT + 6] = &mac_writereg,
+ [RA ... RA + 31] = &mac_writereg,
+ [WUPM ... WUPM + 31] = &mac_writereg,
+ [MTA ... MTA + 127] = &mac_writereg,
+ [VFTA ... VFTA + 127] = &mac_writereg,
+ [FFMT ... FFMT + 254] = &mac_writereg, [FFVT ... FFVT + 254] = &mac_writereg,
+ [PBM ... PBM + 16383] = &mac_writereg,
};
enum { NWRITEOPS = ARRAY_SIZE(macreg_writeops) };
diff --git a/hw/net/e1000e.c b/hw/net/e1000e.c
index 7523e9f5d2..8635ca16c6 100644
--- a/hw/net/e1000e.c
+++ b/hw/net/e1000e.c
@@ -1,37 +1,37 @@
/*
-* QEMU INTEL 82574 GbE NIC emulation
-*
-* Software developer's manuals:
-* http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf
-*
-* Copyright (c) 2015 Ravello Systems LTD (http://ravellosystems.com)
-* Developed by Daynix Computing LTD (http://www.daynix.com)
-*
-* Authors:
-* Dmitry Fleytman <dmitry@daynix.com>
-* Leonid Bloch <leonid@daynix.com>
-* Yan Vugenfirer <yan@daynix.com>
-*
-* Based on work done by:
-* Nir Peleg, Tutis Systems Ltd. for Qumranet Inc.
-* Copyright (c) 2008 Qumranet
-* Based on work done by:
-* Copyright (c) 2007 Dan Aloni
-* Copyright (c) 2004 Antony T Curtis
-*
-* This library is free software; you can redistribute it and/or
-* modify it under the terms of the GNU Lesser General Public
-* License as published by the Free Software Foundation; either
-* version 2.1 of the License, or (at your option) any later version.
-*
-* This library is distributed in the hope that it will be useful,
-* but WITHOUT ANY WARRANTY; without even the implied warranty of
-* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
-* Lesser General Public License for more details.
-*
-* You should have received a copy of the GNU Lesser General Public
-* License along with this library; if not, see <http://www.gnu.org/licenses/>.
-*/
+ * QEMU INTEL 82574 GbE NIC emulation
+ *
+ * Software developer's manuals:
+ * http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf
+ *
+ * Copyright (c) 2015 Ravello Systems LTD (http://ravellosystems.com)
+ * Developed by Daynix Computing LTD (http://www.daynix.com)
+ *
+ * Authors:
+ * Dmitry Fleytman <dmitry@daynix.com>
+ * Leonid Bloch <leonid@daynix.com>
+ * Yan Vugenfirer <yan@daynix.com>
+ *
+ * Based on work done by:
+ * Nir Peleg, Tutis Systems Ltd. for Qumranet Inc.
+ * Copyright (c) 2008 Qumranet
+ * Based on work done by:
+ * Copyright (c) 2007 Dan Aloni
+ * Copyright (c) 2004 Antony T Curtis
+ *
+ * This library is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * This library is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with this library; if not, see <http://www.gnu.org/licenses/>.
+ */
#include "qemu/osdep.h"
#include "qemu/units.h"
@@ -239,9 +239,9 @@ static NetClientInfo net_e1000e_info = {
};
/*
-* EEPROM (NVM) contents documented in Table 36, section 6.1
-* and generally 6.1.2 Software accessed words.
-*/
+ * EEPROM (NVM) contents documented in Table 36, section 6.1
+ * and generally 6.1.2 Software accessed words.
+ */
static const uint16_t e1000e_eeprom_template[64] = {
/* Address | Compat. | ImVer | Compat. */
0x0000, 0x0000, 0x0000, 0x0420, 0xf746, 0x2010, 0xffff, 0xffff,
diff --git a/hw/net/e1000e_core.c b/hw/net/e1000e_core.c
index fc9cdb4528..c4be782558 100644
--- a/hw/net/e1000e_core.c
+++ b/hw/net/e1000e_core.c
@@ -1,37 +1,37 @@
/*
-* Core code for QEMU e1000e emulation
-*
-* Software developer's manuals:
-* http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf
-*
-* Copyright (c) 2015 Ravello Systems LTD (http://ravellosystems.com)
-* Developed by Daynix Computing LTD (http://www.daynix.com)
-*
-* Authors:
-* Dmitry Fleytman <dmitry@daynix.com>
-* Leonid Bloch <leonid@daynix.com>
-* Yan Vugenfirer <yan@daynix.com>
-*
-* Based on work done by:
-* Nir Peleg, Tutis Systems Ltd. for Qumranet Inc.
-* Copyright (c) 2008 Qumranet
-* Based on work done by:
-* Copyright (c) 2007 Dan Aloni
-* Copyright (c) 2004 Antony T Curtis
-*
-* This library is free software; you can redistribute it and/or
-* modify it under the terms of the GNU Lesser General Public
-* License as published by the Free Software Foundation; either
-* version 2.1 of the License, or (at your option) any later version.
-*
-* This library is distributed in the hope that it will be useful,
-* but WITHOUT ANY WARRANTY; without even the implied warranty of
-* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
-* Lesser General Public License for more details.
-*
-* You should have received a copy of the GNU Lesser General Public
-* License along with this library; if not, see <http://www.gnu.org/licenses/>.
-*/
+ * Core code for QEMU e1000e emulation
+ *
+ * Software developer's manuals:
+ * http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf
+ *
+ * Copyright (c) 2015 Ravello Systems LTD (http://ravellosystems.com)
+ * Developed by Daynix Computing LTD (http://www.daynix.com)
+ *
+ * Authors:
+ * Dmitry Fleytman <dmitry@daynix.com>
+ * Leonid Bloch <leonid@daynix.com>
+ * Yan Vugenfirer <yan@daynix.com>
+ *
+ * Based on work done by:
+ * Nir Peleg, Tutis Systems Ltd. for Qumranet Inc.
+ * Copyright (c) 2008 Qumranet
+ * Based on work done by:
+ * Copyright (c) 2007 Dan Aloni
+ * Copyright (c) 2004 Antony T Curtis
+ *
+ * This library is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * This library is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with this library; if not, see <http://www.gnu.org/licenses/>.
+ */
#include "qemu/osdep.h"
#include "qemu/log.h"
@@ -49,8 +49,9 @@
#include "trace.h"
-#define E1000E_MIN_XITR (500) /* No more then 7813 interrupts per
- second according to spec 10.2.4.2 */
+/* No more then 7813 interrupts per second according to spec 10.2.4.2 */
+#define E1000E_MIN_XITR (500)
+
#define E1000E_MAX_TX_FRAGS (64)
static inline void
@@ -282,14 +283,18 @@ e1000e_intrmgr_delay_rx_causes(E1000ECore *core, uint32_t *causes)
core->delayed_causes |= *causes & delayable_causes;
*causes &= ~delayable_causes;
- /* Check if delayed RX interrupts disabled by client
- or if there are causes that cannot be delayed */
+ /*
+ * Check if delayed RX interrupts disabled by client
+ * or if there are causes that cannot be delayed
+ */
if ((rdtr == 0) || (*causes != 0)) {
return false;
}
- /* Check if delayed RX ACK interrupts disabled by client
- and there is an ACK packet received */
+ /*
+ * Check if delayed RX ACK interrupts disabled by client
+ * and there is an ACK packet received
+ */
if ((raid == 0) && (core->delayed_causes & E1000_ICR_ACK)) {
return false;
}
@@ -2515,7 +2520,8 @@ e1000e_set_icr(E1000ECore *core, int index, uint32_t val)
}
icr = core->mac[ICR] & ~val;
- /* Windows driver expects that the "receive overrun" bit and other
+ /*
+ * Windows driver expects that the "receive overrun" bit and other
* ones to be cleared when the "Other" bit (#24) is cleared.
*/
icr = (val & E1000_ICR_OTHER) ? (icr & ~E1000_ICR_OTHER_CAUSES) : icr;
@@ -3269,10 +3275,12 @@ enum { E1000E_NWRITEOPS = ARRAY_SIZE(e1000e_macreg_writeops) };
enum { MAC_ACCESS_PARTIAL = 1 };
-/* The array below combines alias offsets of the index values for the
+/*
+ * The array below combines alias offsets of the index values for the
* MAC registers that have aliases, with the indication of not fully
* implemented registers (lowest bit). This combination is possible
- * because all of the offsets are even. */
+ * because all of the offsets are even.
+ */
static const uint16_t mac_reg_access[E1000E_MAC_SIZE] = {
/* Alias index offsets */
[FCRTL_A] = 0x07fe, [FCRTH_A] = 0x0802,
@@ -3536,10 +3544,10 @@ void e1000e_core_pre_save(E1000ECore *core)
NetClientState *nc = qemu_get_queue(core->owner_nic);
/*
- * If link is down and auto-negotiation is supported and ongoing,
- * complete auto-negotiation immediately. This allows us to look
- * at MII_SR_AUTONEG_COMPLETE to infer link status on load.
- */
+ * If link is down and auto-negotiation is supported and ongoing,
+ * complete auto-negotiation immediately. This allows us to look
+ * at MII_SR_AUTONEG_COMPLETE to infer link status on load.
+ */
if (nc->link_down && e1000e_have_autoneg(core)) {
core->phy[0][PHY_STATUS] |= MII_SR_AUTONEG_COMPLETE;
e1000e_update_flowctl_status(core);
@@ -3557,7 +3565,8 @@ e1000e_core_post_load(E1000ECore *core)
{
NetClientState *nc = qemu_get_queue(core->owner_nic);
- /* nc.link_down can't be migrated, so infer link_down according
+ /*
+ * nc.link_down can't be migrated, so infer link_down according
* to link status bit in core.mac[STATUS].
*/
nc->link_down = (core->mac[STATUS] & E1000_STATUS_LU) == 0;
diff --git a/hw/net/e1000e_core.h b/hw/net/e1000e_core.h
index 4ddb4d2c39..b8f38c47a0 100644
--- a/hw/net/e1000e_core.h
+++ b/hw/net/e1000e_core.h
@@ -1,37 +1,37 @@
/*
-* Core code for QEMU e1000e emulation
-*
-* Software developer's manuals:
-* http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf
-*
-* Copyright (c) 2015 Ravello Systems LTD (http://ravellosystems.com)
-* Developed by Daynix Computing LTD (http://www.daynix.com)
-*
-* Authors:
-* Dmitry Fleytman <dmitry@daynix.com>
-* Leonid Bloch <leonid@daynix.com>
-* Yan Vugenfirer <yan@daynix.com>
-*
-* Based on work done by:
-* Nir Peleg, Tutis Systems Ltd. for Qumranet Inc.
-* Copyright (c) 2008 Qumranet
-* Based on work done by:
-* Copyright (c) 2007 Dan Aloni
-* Copyright (c) 2004 Antony T Curtis
-*
-* This library is free software; you can redistribute it and/or
-* modify it under the terms of the GNU Lesser General Public
-* License as published by the Free Software Foundation; either
-* version 2.1 of the License, or (at your option) any later version.
-*
-* This library is distributed in the hope that it will be useful,
-* but WITHOUT ANY WARRANTY; without even the implied warranty of
-* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
-* Lesser General Public License for more details.
-*
-* You should have received a copy of the GNU Lesser General Public
-* License along with this library; if not, see <http://www.gnu.org/licenses/>.
-*/
+ * Core code for QEMU e1000e emulation
+ *
+ * Software developer's manuals:
+ * http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf
+ *
+ * Copyright (c) 2015 Ravello Systems LTD (http://ravellosystems.com)
+ * Developed by Daynix Computing LTD (http://www.daynix.com)
+ *
+ * Authors:
+ * Dmitry Fleytman <dmitry@daynix.com>
+ * Leonid Bloch <leonid@daynix.com>
+ * Yan Vugenfirer <yan@daynix.com>
+ *
+ * Based on work done by:
+ * Nir Peleg, Tutis Systems Ltd. for Qumranet Inc.
+ * Copyright (c) 2008 Qumranet
+ * Based on work done by:
+ * Copyright (c) 2007 Dan Aloni
+ * Copyright (c) 2004 Antony T Curtis
+ *
+ * This library is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * This library is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with this library; if not, see <http://www.gnu.org/licenses/>.
+ */
#ifndef HW_NET_E1000E_CORE_H
#define HW_NET_E1000E_CORE_H
diff --git a/hw/net/e1000x_common.h b/hw/net/e1000x_common.h
index b7742775c4..bc1ee75b04 100644
--- a/hw/net/e1000x_common.h
+++ b/hw/net/e1000x_common.h
@@ -1,26 +1,26 @@
/*
-* QEMU e1000(e) emulation - shared code
-*
-* Copyright (c) 2008 Qumranet
-*
-* Based on work done by:
-* Nir Peleg, Tutis Systems Ltd. for Qumranet Inc.
-* Copyright (c) 2007 Dan Aloni
-* Copyright (c) 2004 Antony T Curtis
-*
-* This library is free software; you can redistribute it and/or
-* modify it under the terms of the GNU Lesser General Public
-* License as published by the Free Software Foundation; either
-* version 2.1 of the License, or (at your option) any later version.
-*
-* This library is distributed in the hope that it will be useful,
-* but WITHOUT ANY WARRANTY; without even the implied warranty of
-* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
-* Lesser General Public License for more details.
-*
-* You should have received a copy of the GNU Lesser General Public
-* License along with this library; if not, see <http://www.gnu.org/licenses/>.
-*/
+ * QEMU e1000(e) emulation - shared code
+ *
+ * Copyright (c) 2008 Qumranet
+ *
+ * Based on work done by:
+ * Nir Peleg, Tutis Systems Ltd. for Qumranet Inc.
+ * Copyright (c) 2007 Dan Aloni
+ * Copyright (c) 2004 Antony T Curtis
+ *
+ * This library is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * This library is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with this library; if not, see <http://www.gnu.org/licenses/>.
+ */
#ifndef HW_NET_E1000X_COMMON_H
#define HW_NET_E1000X_COMMON_H
--
2.39.1
^ permalink raw reply related [flat|nested] 33+ messages in thread* [PATCH v4 02/28] hw/net: Add more MII definitions
2023-01-30 13:46 [PATCH v4 00/28] e1000x cleanups (preliminary for IGB) Akihiko Odaki
2023-01-30 13:46 ` [PATCH v4 01/28] e1000e: Fix the code style Akihiko Odaki
@ 2023-01-30 13:46 ` Akihiko Odaki
2023-01-30 13:46 ` [PATCH v4 03/28] fsl_etsec: Use hw/net/mii.h Akihiko Odaki
` (25 subsequent siblings)
27 siblings, 0 replies; 33+ messages in thread
From: Akihiko Odaki @ 2023-01-30 13:46 UTC (permalink / raw)
Cc: Jason Wang, Dmitry Fleytman, Thomas Huth, Laurent Vivier,
Paolo Bonzini, qemu-devel, qemu-ppc, Michael S . Tsirkin,
Yan Vugenfirer, Yuri Benditovich, Sriram Yagnaraman,
Alexander Bulekov, Akihiko Odaki
The definitions will be used by igb.
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
---
include/hw/net/mii.h | 13 ++++++++++++-
1 file changed, 12 insertions(+), 1 deletion(-)
diff --git a/include/hw/net/mii.h b/include/hw/net/mii.h
index 4ae4dcce7e..c6a767a49a 100644
--- a/include/hw/net/mii.h
+++ b/include/hw/net/mii.h
@@ -81,20 +81,31 @@
#define MII_ANLPAR_ACK (1 << 14)
#define MII_ANLPAR_PAUSEASY (1 << 11) /* can pause asymmetrically */
#define MII_ANLPAR_PAUSE (1 << 10) /* can pause */
+#define MII_ANLPAR_T4 (1 << 9)
#define MII_ANLPAR_TXFD (1 << 8)
#define MII_ANLPAR_TX (1 << 7)
#define MII_ANLPAR_10FD (1 << 6)
#define MII_ANLPAR_10 (1 << 5)
#define MII_ANLPAR_CSMACD (1 << 0)
-#define MII_ANER_NWAY (1 << 0) /* Can do N-way auto-nego */
+#define MII_ANER_NP (1 << 2) /* Next Page Able */
+#define MII_ANER_NWAY (1 << 0) /* Can do N-way auto-nego */
+#define MII_ANNP_MP (1 << 13) /* Message Page */
+
+#define MII_CTRL1000_MASTER (1 << 11) /* MASTER-SLAVE Manual Configuration Value */
+#define MII_CTRL1000_PORT (1 << 10) /* T2_Repeater/DTE bit */
#define MII_CTRL1000_FULL (1 << 9) /* 1000BASE-T full duplex */
#define MII_CTRL1000_HALF (1 << 8) /* 1000BASE-T half duplex */
+#define MII_STAT1000_LOK (1 << 13) /* Local Receiver Status */
+#define MII_STAT1000_ROK (1 << 12) /* Remote Receiver Status */
#define MII_STAT1000_FULL (1 << 11) /* 1000BASE-T full duplex */
#define MII_STAT1000_HALF (1 << 10) /* 1000BASE-T half duplex */
+#define MII_EXTSTAT_1000T_FD (1 << 13) /* 1000BASE-T Full Duplex */
+#define MII_EXTSTAT_1000T_HD (1 << 12) /* 1000BASE-T Half Duplex */
+
/* List of vendor identifiers */
/* RealTek 8201 */
#define RTL8201CP_PHYID1 0x0000
--
2.39.1
^ permalink raw reply related [flat|nested] 33+ messages in thread* [PATCH v4 03/28] fsl_etsec: Use hw/net/mii.h
2023-01-30 13:46 [PATCH v4 00/28] e1000x cleanups (preliminary for IGB) Akihiko Odaki
2023-01-30 13:46 ` [PATCH v4 01/28] e1000e: Fix the code style Akihiko Odaki
2023-01-30 13:46 ` [PATCH v4 02/28] hw/net: Add more MII definitions Akihiko Odaki
@ 2023-01-30 13:46 ` Akihiko Odaki
2023-01-30 13:46 ` [PATCH v4 04/28] e1000: " Akihiko Odaki
` (24 subsequent siblings)
27 siblings, 0 replies; 33+ messages in thread
From: Akihiko Odaki @ 2023-01-30 13:46 UTC (permalink / raw)
Cc: Jason Wang, Dmitry Fleytman, Thomas Huth, Laurent Vivier,
Paolo Bonzini, qemu-devel, qemu-ppc, Michael S . Tsirkin,
Yan Vugenfirer, Yuri Benditovich, Sriram Yagnaraman,
Alexander Bulekov, Akihiko Odaki, Philippe Mathieu-Daudé
hw/net/mii.h provides common definitions for MII.
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
---
hw/net/fsl_etsec/etsec.c | 11 ++++++-----
hw/net/fsl_etsec/etsec.h | 17 -----------------
hw/net/fsl_etsec/miim.c | 5 +++--
include/hw/net/mii.h | 1 +
4 files changed, 10 insertions(+), 24 deletions(-)
diff --git a/hw/net/fsl_etsec/etsec.c b/hw/net/fsl_etsec/etsec.c
index c753bfb3a8..798ea33d08 100644
--- a/hw/net/fsl_etsec/etsec.c
+++ b/hw/net/fsl_etsec/etsec.c
@@ -29,6 +29,7 @@
#include "qemu/osdep.h"
#include "hw/sysbus.h"
#include "hw/irq.h"
+#include "hw/net/mii.h"
#include "hw/ptimer.h"
#include "hw/qdev-properties.h"
#include "etsec.h"
@@ -339,11 +340,11 @@ static void etsec_reset(DeviceState *d)
etsec->rx_buffer_len = 0;
etsec->phy_status =
- MII_SR_EXTENDED_CAPS | MII_SR_LINK_STATUS | MII_SR_AUTONEG_CAPS |
- MII_SR_AUTONEG_COMPLETE | MII_SR_PREAMBLE_SUPPRESS |
- MII_SR_EXTENDED_STATUS | MII_SR_100T2_HD_CAPS | MII_SR_100T2_FD_CAPS |
- MII_SR_10T_HD_CAPS | MII_SR_10T_FD_CAPS | MII_SR_100X_HD_CAPS |
- MII_SR_100X_FD_CAPS | MII_SR_100T4_CAPS;
+ MII_BMSR_EXTCAP | MII_BMSR_LINK_ST | MII_BMSR_AUTONEG |
+ MII_BMSR_AN_COMP | MII_BMSR_MFPS | MII_BMSR_EXTSTAT |
+ MII_BMSR_100T2_HD | MII_BMSR_100T2_FD |
+ MII_BMSR_10T_HD | MII_BMSR_10T_FD |
+ MII_BMSR_100TX_HD | MII_BMSR_100TX_FD | MII_BMSR_100T4;
etsec_update_irq(etsec);
}
diff --git a/hw/net/fsl_etsec/etsec.h b/hw/net/fsl_etsec/etsec.h
index 3c625c955c..3860864a3f 100644
--- a/hw/net/fsl_etsec/etsec.h
+++ b/hw/net/fsl_etsec/etsec.h
@@ -76,23 +76,6 @@ typedef struct eTSEC_rxtx_bd {
#define FCB_TX_CTU (1 << 1)
#define FCB_TX_NPH (1 << 0)
-/* PHY Status Register */
-#define MII_SR_EXTENDED_CAPS 0x0001 /* Extended register capabilities */
-#define MII_SR_JABBER_DETECT 0x0002 /* Jabber Detected */
-#define MII_SR_LINK_STATUS 0x0004 /* Link Status 1 = link */
-#define MII_SR_AUTONEG_CAPS 0x0008 /* Auto Neg Capable */
-#define MII_SR_REMOTE_FAULT 0x0010 /* Remote Fault Detect */
-#define MII_SR_AUTONEG_COMPLETE 0x0020 /* Auto Neg Complete */
-#define MII_SR_PREAMBLE_SUPPRESS 0x0040 /* Preamble may be suppressed */
-#define MII_SR_EXTENDED_STATUS 0x0100 /* Ext. status info in Reg 0x0F */
-#define MII_SR_100T2_HD_CAPS 0x0200 /* 100T2 Half Duplex Capable */
-#define MII_SR_100T2_FD_CAPS 0x0400 /* 100T2 Full Duplex Capable */
-#define MII_SR_10T_HD_CAPS 0x0800 /* 10T Half Duplex Capable */
-#define MII_SR_10T_FD_CAPS 0x1000 /* 10T Full Duplex Capable */
-#define MII_SR_100X_HD_CAPS 0x2000 /* 100X Half Duplex Capable */
-#define MII_SR_100X_FD_CAPS 0x4000 /* 100X Full Duplex Capable */
-#define MII_SR_100T4_CAPS 0x8000 /* 100T4 Capable */
-
/* eTSEC */
/* Number of register in the device */
diff --git a/hw/net/fsl_etsec/miim.c b/hw/net/fsl_etsec/miim.c
index 6bba01c82a..b48d2cb57b 100644
--- a/hw/net/fsl_etsec/miim.c
+++ b/hw/net/fsl_etsec/miim.c
@@ -23,6 +23,7 @@
*/
#include "qemu/osdep.h"
+#include "hw/net/mii.h"
#include "etsec.h"
#include "registers.h"
@@ -140,8 +141,8 @@ void etsec_miim_link_status(eTSEC *etsec, NetClientState *nc)
{
/* Set link status */
if (nc->link_down) {
- etsec->phy_status &= ~MII_SR_LINK_STATUS;
+ etsec->phy_status &= ~MII_BMSR_LINK_ST;
} else {
- etsec->phy_status |= MII_SR_LINK_STATUS;
+ etsec->phy_status |= MII_BMSR_LINK_ST;
}
}
diff --git a/include/hw/net/mii.h b/include/hw/net/mii.h
index c6a767a49a..ed1bb52b0f 100644
--- a/include/hw/net/mii.h
+++ b/include/hw/net/mii.h
@@ -55,6 +55,7 @@
#define MII_BMCR_CTST (1 << 7) /* Collision test */
#define MII_BMCR_SPEED1000 (1 << 6) /* MSB of Speed (1000) */
+#define MII_BMSR_100T4 (1 << 15) /* Can do 100mbps T4 */
#define MII_BMSR_100TX_FD (1 << 14) /* Can do 100mbps, full-duplex */
#define MII_BMSR_100TX_HD (1 << 13) /* Can do 100mbps, half-duplex */
#define MII_BMSR_10T_FD (1 << 12) /* Can do 10mbps, full-duplex */
--
2.39.1
^ permalink raw reply related [flat|nested] 33+ messages in thread* [PATCH v4 04/28] e1000: Use hw/net/mii.h
2023-01-30 13:46 [PATCH v4 00/28] e1000x cleanups (preliminary for IGB) Akihiko Odaki
` (2 preceding siblings ...)
2023-01-30 13:46 ` [PATCH v4 03/28] fsl_etsec: Use hw/net/mii.h Akihiko Odaki
@ 2023-01-30 13:46 ` Akihiko Odaki
2023-01-30 13:46 ` [PATCH v4 05/28] e1000: Mask registers when writing Akihiko Odaki
` (23 subsequent siblings)
27 siblings, 0 replies; 33+ messages in thread
From: Akihiko Odaki @ 2023-01-30 13:46 UTC (permalink / raw)
Cc: Jason Wang, Dmitry Fleytman, Thomas Huth, Laurent Vivier,
Paolo Bonzini, qemu-devel, qemu-ppc, Michael S . Tsirkin,
Yan Vugenfirer, Yuri Benditovich, Sriram Yagnaraman,
Alexander Bulekov, Akihiko Odaki, Philippe Mathieu-Daudé
hw/net/mii.h provides common definitions for MII.
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
---
hw/net/e1000.c | 86 ++++++++++++++++++------------------
hw/net/e1000_regs.h | 46 --------------------
hw/net/e1000e.c | 1 +
hw/net/e1000e_core.c | 99 +++++++++++++++++++++---------------------
hw/net/e1000x_common.c | 5 ++-
hw/net/e1000x_common.h | 8 ++--
6 files changed, 101 insertions(+), 144 deletions(-)
diff --git a/hw/net/e1000.c b/hw/net/e1000.c
index 8ee30aa37c..9619a2e481 100644
--- a/hw/net/e1000.c
+++ b/hw/net/e1000.c
@@ -26,6 +26,7 @@
#include "qemu/osdep.h"
+#include "hw/net/mii.h"
#include "hw/pci/pci_device.h"
#include "hw/qdev-properties.h"
#include "migration/vmstate.h"
@@ -181,67 +182,67 @@ e1000_autoneg_done(E1000State *s)
static bool
have_autoneg(E1000State *s)
{
- return chkflag(AUTONEG) && (s->phy_reg[PHY_CTRL] & MII_CR_AUTO_NEG_EN);
+ return chkflag(AUTONEG) && (s->phy_reg[MII_BMCR] & MII_BMCR_AUTOEN);
}
static void
set_phy_ctrl(E1000State *s, int index, uint16_t val)
{
- /* bits 0-5 reserved; MII_CR_[RESTART_AUTO_NEG,RESET] are self clearing */
- s->phy_reg[PHY_CTRL] = val & ~(0x3f |
- MII_CR_RESET |
- MII_CR_RESTART_AUTO_NEG);
+ /* bits 0-5 reserved; MII_BMCR_[ANRESTART,RESET] are self clearing */
+ s->phy_reg[MII_BMCR] = val & ~(0x3f |
+ MII_BMCR_RESET |
+ MII_BMCR_ANRESTART);
/*
* QEMU 1.3 does not support link auto-negotiation emulation, so if we
* migrate during auto negotiation, after migration the link will be
* down.
*/
- if (have_autoneg(s) && (val & MII_CR_RESTART_AUTO_NEG)) {
+ if (have_autoneg(s) && (val & MII_BMCR_ANRESTART)) {
e1000x_restart_autoneg(s->mac_reg, s->phy_reg, s->autoneg_timer);
}
}
static void (*phyreg_writeops[])(E1000State *, int, uint16_t) = {
- [PHY_CTRL] = set_phy_ctrl,
+ [MII_BMCR] = set_phy_ctrl,
};
enum { NPHYWRITEOPS = ARRAY_SIZE(phyreg_writeops) };
enum { PHY_R = 1, PHY_W = 2, PHY_RW = PHY_R | PHY_W };
static const char phy_regcap[0x20] = {
- [PHY_STATUS] = PHY_R, [M88E1000_EXT_PHY_SPEC_CTRL] = PHY_RW,
- [PHY_ID1] = PHY_R, [M88E1000_PHY_SPEC_CTRL] = PHY_RW,
- [PHY_CTRL] = PHY_RW, [PHY_1000T_CTRL] = PHY_RW,
- [PHY_LP_ABILITY] = PHY_R, [PHY_1000T_STATUS] = PHY_R,
- [PHY_AUTONEG_ADV] = PHY_RW, [M88E1000_RX_ERR_CNTR] = PHY_R,
- [PHY_ID2] = PHY_R, [M88E1000_PHY_SPEC_STATUS] = PHY_R,
- [PHY_AUTONEG_EXP] = PHY_R,
+ [MII_BMSR] = PHY_R, [M88E1000_EXT_PHY_SPEC_CTRL] = PHY_RW,
+ [MII_PHYID1] = PHY_R, [M88E1000_PHY_SPEC_CTRL] = PHY_RW,
+ [MII_BMCR] = PHY_RW, [MII_CTRL1000] = PHY_RW,
+ [MII_ANLPAR] = PHY_R, [MII_STAT1000] = PHY_R,
+ [MII_ANAR] = PHY_RW, [M88E1000_RX_ERR_CNTR] = PHY_R,
+ [MII_PHYID2] = PHY_R, [M88E1000_PHY_SPEC_STATUS] = PHY_R,
+ [MII_ANER] = PHY_R,
};
-/* PHY_ID2 documented in 8254x_GBe_SDM.pdf, pp. 250 */
+/* MII_PHYID2 documented in 8254x_GBe_SDM.pdf, pp. 250 */
static const uint16_t phy_reg_init[] = {
- [PHY_CTRL] = MII_CR_SPEED_SELECT_MSB |
- MII_CR_FULL_DUPLEX |
- MII_CR_AUTO_NEG_EN,
-
- [PHY_STATUS] = MII_SR_EXTENDED_CAPS |
- MII_SR_LINK_STATUS | /* link initially up */
- MII_SR_AUTONEG_CAPS |
- /* MII_SR_AUTONEG_COMPLETE: initially NOT completed */
- MII_SR_PREAMBLE_SUPPRESS |
- MII_SR_EXTENDED_STATUS |
- MII_SR_10T_HD_CAPS |
- MII_SR_10T_FD_CAPS |
- MII_SR_100X_HD_CAPS |
- MII_SR_100X_FD_CAPS,
-
- [PHY_ID1] = 0x141,
- /* [PHY_ID2] configured per DevId, from e1000_reset() */
- [PHY_AUTONEG_ADV] = 0xde1,
- [PHY_LP_ABILITY] = 0x1e0,
- [PHY_1000T_CTRL] = 0x0e00,
- [PHY_1000T_STATUS] = 0x3c00,
+ [MII_BMCR] = MII_BMCR_SPEED1000 |
+ MII_BMCR_FD |
+ MII_BMCR_AUTOEN,
+
+ [MII_BMSR] = MII_BMSR_EXTCAP |
+ MII_BMSR_LINK_ST | /* link initially up */
+ MII_BMSR_AUTONEG |
+ /* MII_BMSR_AN_COMP: initially NOT completed */
+ MII_BMSR_MFPS |
+ MII_BMSR_EXTSTAT |
+ MII_BMSR_10T_HD |
+ MII_BMSR_10T_FD |
+ MII_BMSR_100TX_HD |
+ MII_BMSR_100TX_FD,
+
+ [MII_PHYID1] = 0x141,
+ /* [MII_PHYID2] configured per DevId, from e1000_reset() */
+ [MII_ANAR] = 0xde1,
+ [MII_ANLPAR] = 0x1e0,
+ [MII_CTRL1000] = 0x0e00,
+ [MII_STAT1000] = 0x3c00,
[M88E1000_PHY_SPEC_CTRL] = 0x360,
[M88E1000_PHY_SPEC_STATUS] = 0xac00,
[M88E1000_EXT_PHY_SPEC_CTRL] = 0x0d60,
@@ -387,7 +388,7 @@ static void e1000_reset(void *opaque)
d->mit_ide = 0;
memset(d->phy_reg, 0, sizeof d->phy_reg);
memmove(d->phy_reg, phy_reg_init, sizeof phy_reg_init);
- d->phy_reg[PHY_ID2] = edc->phy_id2;
+ d->phy_reg[MII_PHYID2] = edc->phy_id2;
memset(d->mac_reg, 0, sizeof d->mac_reg);
memmove(d->mac_reg, mac_reg_init, sizeof mac_reg_init);
d->rxbuf_min_shift = 1;
@@ -561,7 +562,7 @@ e1000_send_packet(E1000State *s, const uint8_t *buf, int size)
PTC1023, PTC1522 };
NetClientState *nc = qemu_get_queue(s->nic);
- if (s->phy_reg[PHY_CTRL] & MII_CR_LOOPBACK) {
+ if (s->phy_reg[MII_BMCR] & MII_BMCR_LOOPBACK) {
qemu_receive_packet(nc, buf, size);
} else {
qemu_send_packet(nc, buf, size);
@@ -842,7 +843,7 @@ e1000_set_link_status(NetClientState *nc)
e1000x_update_regs_on_link_down(s->mac_reg, s->phy_reg);
} else {
if (have_autoneg(s) &&
- !(s->phy_reg[PHY_STATUS] & MII_SR_AUTONEG_COMPLETE)) {
+ !(s->phy_reg[MII_BMSR] & MII_BMSR_AN_COMP)) {
e1000x_restart_autoneg(s->mac_reg, s->phy_reg, s->autoneg_timer);
} else {
e1000_link_up(s);
@@ -1416,10 +1417,10 @@ static int e1000_pre_save(void *opaque)
/*
* If link is down and auto-negotiation is supported and ongoing,
* complete auto-negotiation immediately. This allows us to look
- * at MII_SR_AUTONEG_COMPLETE to infer link status on load.
+ * at MII_BMSR_AN_COMP to infer link status on load.
*/
if (nc->link_down && have_autoneg(s)) {
- s->phy_reg[PHY_STATUS] |= MII_SR_AUTONEG_COMPLETE;
+ s->phy_reg[MII_BMSR] |= MII_BMSR_AN_COMP;
}
/* Decide which set of props to migrate in the main structure */
@@ -1458,8 +1459,7 @@ static int e1000_post_load(void *opaque, int version_id)
* Alternatively, restart link negotiation if it was in progress. */
nc->link_down = (s->mac_reg[STATUS] & E1000_STATUS_LU) == 0;
- if (have_autoneg(s) &&
- !(s->phy_reg[PHY_STATUS] & MII_SR_AUTONEG_COMPLETE)) {
+ if (have_autoneg(s) && !(s->phy_reg[MII_BMSR] & MII_BMSR_AN_COMP)) {
nc->link_down = false;
timer_mod(s->autoneg_timer,
qemu_clock_get_ms(QEMU_CLOCK_VIRTUAL) + 500);
diff --git a/hw/net/e1000_regs.h b/hw/net/e1000_regs.h
index 59e050742b..59d6eb3189 100644
--- a/hw/net/e1000_regs.h
+++ b/hw/net/e1000_regs.h
@@ -474,20 +474,6 @@
#define E1000_TARC_ENABLE BIT(10)
/* PHY 1000 MII Register/Bit Definitions */
-/* PHY Registers defined by IEEE */
-#define PHY_CTRL 0x00 /* Control Register */
-#define PHY_STATUS 0x01 /* Status Regiser */
-#define PHY_ID1 0x02 /* Phy Id Reg (word 1) */
-#define PHY_ID2 0x03 /* Phy Id Reg (word 2) */
-#define PHY_AUTONEG_ADV 0x04 /* Autoneg Advertisement */
-#define PHY_LP_ABILITY 0x05 /* Link Partner Ability (Base Page) */
-#define PHY_AUTONEG_EXP 0x06 /* Autoneg Expansion Reg */
-#define PHY_NEXT_PAGE_TX 0x07 /* Next Page TX */
-#define PHY_LP_NEXT_PAGE 0x08 /* Link Partner Next Page */
-#define PHY_1000T_CTRL 0x09 /* 1000Base-T Control Reg */
-#define PHY_1000T_STATUS 0x0A /* 1000Base-T Status Reg */
-#define PHY_EXT_STATUS 0x0F /* Extended Status Reg */
-
/* 82574-specific registers */
#define PHY_COPPER_CTRL1 0x10 /* Copper Specific Control Register 1 */
#define PHY_COPPER_STAT1 0x11 /* Copper Specific Status Register 1 */
@@ -539,38 +525,6 @@
#define M88E1000_PHY_VCO_REG_BIT8 0x100 /* Bits 8 & 11 are adjusted for */
#define M88E1000_PHY_VCO_REG_BIT11 0x800 /* improved BER performance */
-/* PHY Control Register */
-#define MII_CR_SPEED_SELECT_MSB 0x0040 /* bits 6,13: 10=1000, 01=100, 00=10 */
-#define MII_CR_COLL_TEST_ENABLE 0x0080 /* Collision test enable */
-#define MII_CR_FULL_DUPLEX 0x0100 /* FDX =1, half duplex =0 */
-#define MII_CR_RESTART_AUTO_NEG 0x0200 /* Restart auto negotiation */
-#define MII_CR_ISOLATE 0x0400 /* Isolate PHY from MII */
-#define MII_CR_POWER_DOWN 0x0800 /* Power down */
-#define MII_CR_AUTO_NEG_EN 0x1000 /* Auto Neg Enable */
-#define MII_CR_SPEED_SELECT_LSB 0x2000 /* bits 6,13: 10=1000, 01=100, 00=10 */
-#define MII_CR_LOOPBACK 0x4000 /* 0 = normal, 1 = loopback */
-#define MII_CR_RESET 0x8000 /* 0 = normal, 1 = PHY reset */
-
-/* PHY Status Register */
-#define MII_SR_EXTENDED_CAPS 0x0001 /* Extended register capabilities */
-#define MII_SR_JABBER_DETECT 0x0002 /* Jabber Detected */
-#define MII_SR_LINK_STATUS 0x0004 /* Link Status 1 = link */
-#define MII_SR_AUTONEG_CAPS 0x0008 /* Auto Neg Capable */
-#define MII_SR_REMOTE_FAULT 0x0010 /* Remote Fault Detect */
-#define MII_SR_AUTONEG_COMPLETE 0x0020 /* Auto Neg Complete */
-#define MII_SR_PREAMBLE_SUPPRESS 0x0040 /* Preamble may be suppressed */
-#define MII_SR_EXTENDED_STATUS 0x0100 /* Ext. status info in Reg 0x0F */
-#define MII_SR_100T2_HD_CAPS 0x0200 /* 100T2 Half Duplex Capable */
-#define MII_SR_100T2_FD_CAPS 0x0400 /* 100T2 Full Duplex Capable */
-#define MII_SR_10T_HD_CAPS 0x0800 /* 10T Half Duplex Capable */
-#define MII_SR_10T_FD_CAPS 0x1000 /* 10T Full Duplex Capable */
-#define MII_SR_100X_HD_CAPS 0x2000 /* 100X Half Duplex Capable */
-#define MII_SR_100X_FD_CAPS 0x4000 /* 100X Full Duplex Capable */
-#define MII_SR_100T4_CAPS 0x8000 /* 100T4 Capable */
-
-/* PHY Link Partner Ability Register */
-#define MII_LPAR_LPACK 0x4000 /* Acked by link partner */
-
/* Interrupt Cause Read */
#define E1000_ICR_TXDW 0x00000001 /* Transmit desc written back */
#define E1000_ICR_TXQE 0x00000002 /* Transmit Queue empty */
diff --git a/hw/net/e1000e.c b/hw/net/e1000e.c
index 8635ca16c6..d591d01c07 100644
--- a/hw/net/e1000e.c
+++ b/hw/net/e1000e.c
@@ -42,6 +42,7 @@
#include "qemu/range.h"
#include "sysemu/sysemu.h"
#include "hw/hw.h"
+#include "hw/net/mii.h"
#include "hw/pci/msi.h"
#include "hw/pci/msix.h"
#include "hw/qdev-properties.h"
diff --git a/hw/net/e1000e_core.c b/hw/net/e1000e_core.c
index c4be782558..181c1e0c2a 100644
--- a/hw/net/e1000e_core.c
+++ b/hw/net/e1000e_core.c
@@ -37,6 +37,7 @@
#include "qemu/log.h"
#include "net/net.h"
#include "net/tap.h"
+#include "hw/net/mii.h"
#include "hw/pci/msi.h"
#include "hw/pci/msix.h"
#include "sysemu/runstate.h"
@@ -659,7 +660,7 @@ e1000e_tx_pkt_send(E1000ECore *core, struct e1000e_tx *tx, int queue_index)
net_tx_pkt_dump(tx->tx_pkt);
- if ((core->phy[0][PHY_CTRL] & MII_CR_LOOPBACK) ||
+ if ((core->phy[0][MII_BMCR] & MII_BMCR_LOOPBACK) ||
((core->mac[RCTL] & E1000_RCTL_LBM_MAC) == E1000_RCTL_LBM_MAC)) {
return net_tx_pkt_send_loopback(tx->tx_pkt, queue);
} else {
@@ -1797,13 +1798,13 @@ e1000e_receive_iov(E1000ECore *core, const struct iovec *iov, int iovcnt)
static inline bool
e1000e_have_autoneg(E1000ECore *core)
{
- return core->phy[0][PHY_CTRL] & MII_CR_AUTO_NEG_EN;
+ return core->phy[0][MII_BMCR] & MII_BMCR_AUTOEN;
}
static void e1000e_update_flowctl_status(E1000ECore *core)
{
if (e1000e_have_autoneg(core) &&
- core->phy[0][PHY_STATUS] & MII_SR_AUTONEG_COMPLETE) {
+ core->phy[0][MII_BMSR] & MII_BMSR_AN_COMP) {
trace_e1000e_link_autoneg_flowctl(true);
core->mac[CTRL] |= E1000_CTRL_TFCE | E1000_CTRL_RFCE;
} else {
@@ -1821,12 +1822,12 @@ e1000e_link_down(E1000ECore *core)
static inline void
e1000e_set_phy_ctrl(E1000ECore *core, int index, uint16_t val)
{
- /* bits 0-5 reserved; MII_CR_[RESTART_AUTO_NEG,RESET] are self clearing */
- core->phy[0][PHY_CTRL] = val & ~(0x3f |
- MII_CR_RESET |
- MII_CR_RESTART_AUTO_NEG);
+ /* bits 0-5 reserved; MII_BMCR_[ANRESTART,RESET] are self clearing */
+ core->phy[0][MII_BMCR] = val & ~(0x3f |
+ MII_BMCR_RESET |
+ MII_BMCR_ANRESTART);
- if ((val & MII_CR_RESTART_AUTO_NEG) &&
+ if ((val & MII_BMCR_ANRESTART) &&
e1000e_have_autoneg(core)) {
e1000x_restart_autoneg(core->mac, core->phy[0], core->autoneg_timer);
}
@@ -1860,7 +1861,7 @@ e1000e_core_set_link_status(E1000ECore *core)
e1000x_update_regs_on_link_down(core->mac, core->phy[0]);
} else {
if (e1000e_have_autoneg(core) &&
- !(core->phy[0][PHY_STATUS] & MII_SR_AUTONEG_COMPLETE)) {
+ !(core->phy[0][MII_BMSR] & MII_BMSR_AN_COMP)) {
e1000x_restart_autoneg(core->mac, core->phy[0],
core->autoneg_timer);
} else {
@@ -2002,7 +2003,7 @@ static
void(*e1000e_phyreg_writeops[E1000E_PHY_PAGES][E1000E_PHY_PAGE_SIZE])
(E1000ECore *, int, uint16_t) = {
[0] = {
- [PHY_CTRL] = e1000e_set_phy_ctrl,
+ [MII_BMCR] = e1000e_set_phy_ctrl,
[PHY_PAGE] = e1000e_set_phy_page,
[PHY_OEM_BITS] = e1000e_set_phy_oem_bits
}
@@ -2274,19 +2275,19 @@ e1000e_get_reg_index_with_offset(const uint16_t *mac_reg_access, hwaddr addr)
static const char e1000e_phy_regcap[E1000E_PHY_PAGES][0x20] = {
[0] = {
- [PHY_CTRL] = PHY_ANYPAGE | PHY_RW,
- [PHY_STATUS] = PHY_ANYPAGE | PHY_R,
- [PHY_ID1] = PHY_ANYPAGE | PHY_R,
- [PHY_ID2] = PHY_ANYPAGE | PHY_R,
- [PHY_AUTONEG_ADV] = PHY_ANYPAGE | PHY_RW,
- [PHY_LP_ABILITY] = PHY_ANYPAGE | PHY_R,
- [PHY_AUTONEG_EXP] = PHY_ANYPAGE | PHY_R,
- [PHY_NEXT_PAGE_TX] = PHY_ANYPAGE | PHY_RW,
- [PHY_LP_NEXT_PAGE] = PHY_ANYPAGE | PHY_R,
- [PHY_1000T_CTRL] = PHY_ANYPAGE | PHY_RW,
- [PHY_1000T_STATUS] = PHY_ANYPAGE | PHY_R,
- [PHY_EXT_STATUS] = PHY_ANYPAGE | PHY_R,
- [PHY_PAGE] = PHY_ANYPAGE | PHY_RW,
+ [MII_BMCR] = PHY_ANYPAGE | PHY_RW,
+ [MII_BMSR] = PHY_ANYPAGE | PHY_R,
+ [MII_PHYID1] = PHY_ANYPAGE | PHY_R,
+ [MII_PHYID2] = PHY_ANYPAGE | PHY_R,
+ [MII_ANAR] = PHY_ANYPAGE | PHY_RW,
+ [MII_ANLPAR] = PHY_ANYPAGE | PHY_R,
+ [MII_ANER] = PHY_ANYPAGE | PHY_R,
+ [MII_ANNP] = PHY_ANYPAGE | PHY_RW,
+ [MII_ANLPRNP] = PHY_ANYPAGE | PHY_R,
+ [MII_CTRL1000] = PHY_ANYPAGE | PHY_RW,
+ [MII_STAT1000] = PHY_ANYPAGE | PHY_R,
+ [MII_EXTSTAT] = PHY_ANYPAGE | PHY_R,
+ [PHY_PAGE] = PHY_ANYPAGE | PHY_RW,
[PHY_COPPER_CTRL1] = PHY_RW,
[PHY_COPPER_STAT1] = PHY_R,
@@ -3355,7 +3356,7 @@ static void
e1000e_autoneg_resume(E1000ECore *core)
{
if (e1000e_have_autoneg(core) &&
- !(core->phy[0][PHY_STATUS] & MII_SR_AUTONEG_COMPLETE)) {
+ !(core->phy[0][MII_BMSR] & MII_BMSR_AN_COMP)) {
qemu_get_queue(core->owner_nic)->link_down = false;
timer_mod(core->autoneg_timer,
qemu_clock_get_ms(QEMU_CLOCK_VIRTUAL) + 500);
@@ -3430,29 +3431,29 @@ e1000e_core_pci_uninit(E1000ECore *core)
static const uint16_t
e1000e_phy_reg_init[E1000E_PHY_PAGES][E1000E_PHY_PAGE_SIZE] = {
[0] = {
- [PHY_CTRL] = MII_CR_SPEED_SELECT_MSB |
- MII_CR_FULL_DUPLEX |
- MII_CR_AUTO_NEG_EN,
-
- [PHY_STATUS] = MII_SR_EXTENDED_CAPS |
- MII_SR_LINK_STATUS |
- MII_SR_AUTONEG_CAPS |
- MII_SR_PREAMBLE_SUPPRESS |
- MII_SR_EXTENDED_STATUS |
- MII_SR_10T_HD_CAPS |
- MII_SR_10T_FD_CAPS |
- MII_SR_100X_HD_CAPS |
- MII_SR_100X_FD_CAPS,
-
- [PHY_ID1] = 0x141,
- [PHY_ID2] = E1000_PHY_ID2_82574x,
- [PHY_AUTONEG_ADV] = 0xde1,
- [PHY_LP_ABILITY] = 0x7e0,
- [PHY_AUTONEG_EXP] = BIT(2),
- [PHY_NEXT_PAGE_TX] = BIT(0) | BIT(13),
- [PHY_1000T_CTRL] = BIT(8) | BIT(9) | BIT(10) | BIT(11),
- [PHY_1000T_STATUS] = 0x3c00,
- [PHY_EXT_STATUS] = BIT(12) | BIT(13),
+ [MII_BMCR] = MII_BMCR_SPEED1000 |
+ MII_BMCR_FD |
+ MII_BMCR_AUTOEN,
+
+ [MII_BMSR] = MII_BMSR_EXTCAP |
+ MII_BMSR_LINK_ST |
+ MII_BMSR_AUTONEG |
+ MII_BMSR_MFPS |
+ MII_BMSR_EXTSTAT |
+ MII_BMSR_10T_HD |
+ MII_BMSR_10T_FD |
+ MII_BMSR_100TX_HD |
+ MII_BMSR_100TX_FD,
+
+ [MII_PHYID1] = 0x141,
+ [MII_PHYID2] = E1000_PHY_ID2_82574x,
+ [MII_ANAR] = 0xde1,
+ [MII_ANLPAR] = 0x7e0,
+ [MII_ANER] = BIT(2),
+ [MII_ANNP] = BIT(0) | BIT(13),
+ [MII_CTRL1000] = BIT(8) | BIT(9) | BIT(10) | BIT(11),
+ [MII_STAT1000] = 0x3c00,
+ [MII_EXTSTAT] = BIT(12) | BIT(13),
[PHY_COPPER_CTRL1] = BIT(5) | BIT(6) | BIT(8) | BIT(9) |
BIT(12) | BIT(13),
@@ -3546,10 +3547,10 @@ void e1000e_core_pre_save(E1000ECore *core)
/*
* If link is down and auto-negotiation is supported and ongoing,
* complete auto-negotiation immediately. This allows us to look
- * at MII_SR_AUTONEG_COMPLETE to infer link status on load.
+ * at MII_BMSR_AN_COMP to infer link status on load.
*/
if (nc->link_down && e1000e_have_autoneg(core)) {
- core->phy[0][PHY_STATUS] |= MII_SR_AUTONEG_COMPLETE;
+ core->phy[0][MII_BMSR] |= MII_BMSR_AN_COMP;
e1000e_update_flowctl_status(core);
}
diff --git a/hw/net/e1000x_common.c b/hw/net/e1000x_common.c
index 2f43e8cd13..e79d4c79bd 100644
--- a/hw/net/e1000x_common.c
+++ b/hw/net/e1000x_common.c
@@ -24,6 +24,7 @@
#include "qemu/osdep.h"
#include "qemu/units.h"
+#include "hw/net/mii.h"
#include "hw/pci/pci_device.h"
#include "net/net.h"
@@ -152,8 +153,8 @@ void e1000x_reset_mac_addr(NICState *nic, uint32_t *mac_regs,
void e1000x_update_regs_on_autoneg_done(uint32_t *mac, uint16_t *phy)
{
e1000x_update_regs_on_link_up(mac, phy);
- phy[PHY_LP_ABILITY] |= MII_LPAR_LPACK;
- phy[PHY_STATUS] |= MII_SR_AUTONEG_COMPLETE;
+ phy[MII_ANLPAR] |= MII_ANLPAR_ACK;
+ phy[MII_BMSR] |= MII_BMSR_AN_COMP;
trace_e1000x_link_negotiation_done();
}
diff --git a/hw/net/e1000x_common.h b/hw/net/e1000x_common.h
index bc1ee75b04..3501e4855a 100644
--- a/hw/net/e1000x_common.h
+++ b/hw/net/e1000x_common.h
@@ -152,16 +152,16 @@ static inline void
e1000x_update_regs_on_link_down(uint32_t *mac, uint16_t *phy)
{
mac[STATUS] &= ~E1000_STATUS_LU;
- phy[PHY_STATUS] &= ~MII_SR_LINK_STATUS;
- phy[PHY_STATUS] &= ~MII_SR_AUTONEG_COMPLETE;
- phy[PHY_LP_ABILITY] &= ~MII_LPAR_LPACK;
+ phy[MII_BMSR] &= ~MII_BMSR_LINK_ST;
+ phy[MII_BMSR] &= ~MII_BMSR_AN_COMP;
+ phy[MII_ANLPAR] &= ~MII_ANLPAR_ACK;
}
static inline void
e1000x_update_regs_on_link_up(uint32_t *mac, uint16_t *phy)
{
mac[STATUS] |= E1000_STATUS_LU;
- phy[PHY_STATUS] |= MII_SR_LINK_STATUS;
+ phy[MII_BMSR] |= MII_BMSR_LINK_ST;
}
void e1000x_update_rx_total_stats(uint32_t *mac,
--
2.39.1
^ permalink raw reply related [flat|nested] 33+ messages in thread* [PATCH v4 05/28] e1000: Mask registers when writing
2023-01-30 13:46 [PATCH v4 00/28] e1000x cleanups (preliminary for IGB) Akihiko Odaki
` (3 preceding siblings ...)
2023-01-30 13:46 ` [PATCH v4 04/28] e1000: " Akihiko Odaki
@ 2023-01-30 13:46 ` Akihiko Odaki
2023-01-30 13:46 ` [PATCH v4 06/28] e1000e: " Akihiko Odaki
` (22 subsequent siblings)
27 siblings, 0 replies; 33+ messages in thread
From: Akihiko Odaki @ 2023-01-30 13:46 UTC (permalink / raw)
Cc: Jason Wang, Dmitry Fleytman, Thomas Huth, Laurent Vivier,
Paolo Bonzini, qemu-devel, qemu-ppc, Michael S . Tsirkin,
Yan Vugenfirer, Yuri Benditovich, Sriram Yagnaraman,
Alexander Bulekov, Akihiko Odaki
When a register has effective bits fewer than their width, the old code
inconsistently masked when writing or reading. Make the code consistent
by always masking when writing, and remove some code duplication.
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
---
hw/net/e1000.c | 84 +++++++++++++++++++-------------------------------
1 file changed, 31 insertions(+), 53 deletions(-)
diff --git a/hw/net/e1000.c b/hw/net/e1000.c
index 9619a2e481..0925a99511 100644
--- a/hw/net/e1000.c
+++ b/hw/net/e1000.c
@@ -1062,30 +1062,6 @@ mac_readreg(E1000State *s, int index)
return s->mac_reg[index];
}
-static uint32_t
-mac_low4_read(E1000State *s, int index)
-{
- return s->mac_reg[index] & 0xf;
-}
-
-static uint32_t
-mac_low11_read(E1000State *s, int index)
-{
- return s->mac_reg[index] & 0x7ff;
-}
-
-static uint32_t
-mac_low13_read(E1000State *s, int index)
-{
- return s->mac_reg[index] & 0x1fff;
-}
-
-static uint32_t
-mac_low16_read(E1000State *s, int index)
-{
- return s->mac_reg[index] & 0xffff;
-}
-
static uint32_t
mac_icr_read(E1000State *s, int index)
{
@@ -1138,11 +1114,17 @@ set_rdt(E1000State *s, int index, uint32_t val)
}
}
-static void
-set_16bit(E1000State *s, int index, uint32_t val)
-{
- s->mac_reg[index] = val & 0xffff;
-}
+#define LOW_BITS_SET_FUNC(num) \
+ static void \
+ set_##num##bit(E1000State *s, int index, uint32_t val) \
+ { \
+ s->mac_reg[index] = val & (BIT(num) - 1); \
+ }
+
+LOW_BITS_SET_FUNC(4)
+LOW_BITS_SET_FUNC(11)
+LOW_BITS_SET_FUNC(13)
+LOW_BITS_SET_FUNC(16)
static void
set_dlen(E1000State *s, int index, uint32_t val)
@@ -1196,7 +1178,9 @@ static const readops macreg_readops[] = {
getreg(XONRXC), getreg(XONTXC), getreg(XOFFRXC), getreg(XOFFTXC),
getreg(RFC), getreg(RJC), getreg(RNBC), getreg(TSCTFC),
getreg(MGTPRC), getreg(MGTPDC), getreg(MGTPTC), getreg(GORCL),
- getreg(GOTCL),
+ getreg(GOTCL), getreg(RDFH), getreg(RDFT), getreg(RDFHS),
+ getreg(RDFTS), getreg(RDFPC), getreg(TDFH), getreg(TDFT),
+ getreg(TDFHS), getreg(TDFTS), getreg(TDFPC), getreg(AIT),
[TOTH] = mac_read_clr8, [TORH] = mac_read_clr8,
[GOTCH] = mac_read_clr8, [GORCH] = mac_read_clr8,
@@ -1214,22 +1198,15 @@ static const readops macreg_readops[] = {
[MPTC] = mac_read_clr4,
[ICR] = mac_icr_read, [EECD] = get_eecd,
[EERD] = flash_eerd_read,
- [RDFH] = mac_low13_read, [RDFT] = mac_low13_read,
- [RDFHS] = mac_low13_read, [RDFTS] = mac_low13_read,
- [RDFPC] = mac_low13_read,
- [TDFH] = mac_low11_read, [TDFT] = mac_low11_read,
- [TDFHS] = mac_low13_read, [TDFTS] = mac_low13_read,
- [TDFPC] = mac_low13_read,
- [AIT] = mac_low16_read,
[CRCERRS ... MPC] = &mac_readreg,
[IP6AT ... IP6AT + 3] = &mac_readreg, [IP4AT ... IP4AT + 6] = &mac_readreg,
- [FFLT ... FFLT + 6] = &mac_low11_read,
+ [FFLT ... FFLT + 6] = &mac_readreg,
[RA ... RA + 31] = &mac_readreg,
[WUPM ... WUPM + 31] = &mac_readreg,
[MTA ... MTA + 127] = &mac_readreg,
[VFTA ... VFTA + 127] = &mac_readreg,
- [FFMT ... FFMT + 254] = &mac_low4_read,
+ [FFMT ... FFMT + 254] = &mac_readreg,
[FFVT ... FFVT + 254] = &mac_readreg,
[PBM ... PBM + 16383] = &mac_readreg,
};
@@ -1241,26 +1218,27 @@ static const writeops macreg_writeops[] = {
putreg(PBA), putreg(EERD), putreg(SWSM), putreg(WUFC),
putreg(TDBAL), putreg(TDBAH), putreg(TXDCTL), putreg(RDBAH),
putreg(RDBAL), putreg(LEDCTL), putreg(VET), putreg(FCRUC),
- putreg(TDFH), putreg(TDFT), putreg(TDFHS), putreg(TDFTS),
- putreg(TDFPC), putreg(RDFH), putreg(RDFT), putreg(RDFHS),
- putreg(RDFTS), putreg(RDFPC), putreg(IPAV), putreg(WUC),
- putreg(WUS), putreg(AIT),
-
- [TDLEN] = set_dlen, [RDLEN] = set_dlen, [TCTL] = set_tctl,
- [TDT] = set_tctl, [MDIC] = set_mdic, [ICS] = set_ics,
- [TDH] = set_16bit, [RDH] = set_16bit, [RDT] = set_rdt,
- [IMC] = set_imc, [IMS] = set_ims, [ICR] = set_icr,
- [EECD] = set_eecd, [RCTL] = set_rx_control, [CTRL] = set_ctrl,
- [RDTR] = set_16bit, [RADV] = set_16bit, [TADV] = set_16bit,
- [ITR] = set_16bit,
+ putreg(IPAV), putreg(WUC),
+ putreg(WUS),
+
+ [TDLEN] = set_dlen, [RDLEN] = set_dlen, [TCTL] = set_tctl,
+ [TDT] = set_tctl, [MDIC] = set_mdic, [ICS] = set_ics,
+ [TDH] = set_16bit, [RDH] = set_16bit, [RDT] = set_rdt,
+ [IMC] = set_imc, [IMS] = set_ims, [ICR] = set_icr,
+ [EECD] = set_eecd, [RCTL] = set_rx_control, [CTRL] = set_ctrl,
+ [RDTR] = set_16bit, [RADV] = set_16bit, [TADV] = set_16bit,
+ [ITR] = set_16bit, [TDFH] = set_11bit, [TDFT] = set_11bit,
+ [TDFHS] = set_13bit, [TDFTS] = set_13bit, [TDFPC] = set_13bit,
+ [RDFH] = set_13bit, [RDFT] = set_13bit, [RDFHS] = set_13bit,
+ [RDFTS] = set_13bit, [RDFPC] = set_13bit, [AIT] = set_16bit,
[IP6AT ... IP6AT + 3] = &mac_writereg, [IP4AT ... IP4AT + 6] = &mac_writereg,
- [FFLT ... FFLT + 6] = &mac_writereg,
+ [FFLT ... FFLT + 6] = &set_11bit,
[RA ... RA + 31] = &mac_writereg,
[WUPM ... WUPM + 31] = &mac_writereg,
[MTA ... MTA + 127] = &mac_writereg,
[VFTA ... VFTA + 127] = &mac_writereg,
- [FFMT ... FFMT + 254] = &mac_writereg, [FFVT ... FFVT + 254] = &mac_writereg,
+ [FFMT ... FFMT + 254] = &set_4bit, [FFVT ... FFVT + 254] = &mac_writereg,
[PBM ... PBM + 16383] = &mac_writereg,
};
--
2.39.1
^ permalink raw reply related [flat|nested] 33+ messages in thread* [PATCH v4 06/28] e1000e: Mask registers when writing
2023-01-30 13:46 [PATCH v4 00/28] e1000x cleanups (preliminary for IGB) Akihiko Odaki
` (4 preceding siblings ...)
2023-01-30 13:46 ` [PATCH v4 05/28] e1000: Mask registers when writing Akihiko Odaki
@ 2023-01-30 13:46 ` Akihiko Odaki
2023-01-30 13:46 ` [PATCH v4 07/28] e1000: Use more constant definitions Akihiko Odaki
` (21 subsequent siblings)
27 siblings, 0 replies; 33+ messages in thread
From: Akihiko Odaki @ 2023-01-30 13:46 UTC (permalink / raw)
Cc: Jason Wang, Dmitry Fleytman, Thomas Huth, Laurent Vivier,
Paolo Bonzini, qemu-devel, qemu-ppc, Michael S . Tsirkin,
Yan Vugenfirer, Yuri Benditovich, Sriram Yagnaraman,
Alexander Bulekov, Akihiko Odaki
When a register has effective bits fewer than their width, the old code
inconsistently masked when writing or reading. Make the code consistent
by always masking when writing, and remove some code duplication.
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
---
hw/net/e1000e_core.c | 94 +++++++++++++++++++-------------------------
1 file changed, 40 insertions(+), 54 deletions(-)
diff --git a/hw/net/e1000e_core.c b/hw/net/e1000e_core.c
index 181c1e0c2a..e6fc85ea51 100644
--- a/hw/net/e1000e_core.c
+++ b/hw/net/e1000e_core.c
@@ -2440,17 +2440,19 @@ e1000e_set_fcrtl(E1000ECore *core, int index, uint32_t val)
core->mac[FCRTL] = val & 0x8000FFF8;
}
-static inline void
-e1000e_set_16bit(E1000ECore *core, int index, uint32_t val)
-{
- core->mac[index] = val & 0xffff;
-}
+#define E1000E_LOW_BITS_SET_FUNC(num) \
+ static void \
+ e1000e_set_##num##bit(E1000ECore *core, int index, uint32_t val) \
+ { \
+ core->mac[index] = val & (BIT(num) - 1); \
+ }
-static void
-e1000e_set_12bit(E1000ECore *core, int index, uint32_t val)
-{
- core->mac[index] = val & 0xfff;
-}
+E1000E_LOW_BITS_SET_FUNC(4)
+E1000E_LOW_BITS_SET_FUNC(6)
+E1000E_LOW_BITS_SET_FUNC(11)
+E1000E_LOW_BITS_SET_FUNC(12)
+E1000E_LOW_BITS_SET_FUNC(13)
+E1000E_LOW_BITS_SET_FUNC(16)
static void
e1000e_set_vet(E1000ECore *core, int index, uint32_t val)
@@ -2621,22 +2623,6 @@ e1000e_mac_ims_read(E1000ECore *core, int index)
return core->mac[IMS];
}
-#define E1000E_LOW_BITS_READ_FUNC(num) \
- static uint32_t \
- e1000e_mac_low##num##_read(E1000ECore *core, int index) \
- { \
- return core->mac[index] & (BIT(num) - 1); \
- } \
-
-#define E1000E_LOW_BITS_READ(num) \
- e1000e_mac_low##num##_read
-
-E1000E_LOW_BITS_READ_FUNC(4);
-E1000E_LOW_BITS_READ_FUNC(6);
-E1000E_LOW_BITS_READ_FUNC(11);
-E1000E_LOW_BITS_READ_FUNC(13);
-E1000E_LOW_BITS_READ_FUNC(16);
-
static uint32_t
e1000e_mac_swsm_read(E1000ECore *core, int index)
{
@@ -2930,7 +2916,19 @@ static const readops e1000e_macreg_readops[] = {
e1000e_getreg(LATECOL),
e1000e_getreg(SEQEC),
e1000e_getreg(XONTXC),
+ e1000e_getreg(AIT),
+ e1000e_getreg(TDFH),
+ e1000e_getreg(TDFT),
+ e1000e_getreg(TDFHS),
+ e1000e_getreg(TDFTS),
+ e1000e_getreg(TDFPC),
e1000e_getreg(WUS),
+ e1000e_getreg(PBS),
+ e1000e_getreg(RDFH),
+ e1000e_getreg(RDFT),
+ e1000e_getreg(RDFHS),
+ e1000e_getreg(RDFTS),
+ e1000e_getreg(RDFPC),
e1000e_getreg(GORCL),
e1000e_getreg(MGTPRC),
e1000e_getreg(EERD),
@@ -3066,16 +3064,9 @@ static const readops e1000e_macreg_readops[] = {
[MPTC] = e1000e_mac_read_clr4,
[IAC] = e1000e_mac_read_clr4,
[ICR] = e1000e_mac_icr_read,
- [RDFH] = E1000E_LOW_BITS_READ(13),
- [RDFHS] = E1000E_LOW_BITS_READ(13),
- [RDFPC] = E1000E_LOW_BITS_READ(13),
- [TDFH] = E1000E_LOW_BITS_READ(13),
- [TDFHS] = E1000E_LOW_BITS_READ(13),
[STATUS] = e1000e_get_status,
[TARC0] = e1000e_get_tarc,
- [PBS] = E1000E_LOW_BITS_READ(6),
[ICS] = e1000e_mac_ics_read,
- [AIT] = E1000E_LOW_BITS_READ(16),
[TORH] = e1000e_mac_read_clr8,
[GORCH] = e1000e_mac_read_clr8,
[PRC127] = e1000e_mac_read_clr4,
@@ -3091,11 +3082,6 @@ static const readops e1000e_macreg_readops[] = {
[BPTC] = e1000e_mac_read_clr4,
[TSCTC] = e1000e_mac_read_clr4,
[ITR] = e1000e_mac_itr_read,
- [RDFT] = E1000E_LOW_BITS_READ(13),
- [RDFTS] = E1000E_LOW_BITS_READ(13),
- [TDFPC] = E1000E_LOW_BITS_READ(13),
- [TDFT] = E1000E_LOW_BITS_READ(13),
- [TDFTS] = E1000E_LOW_BITS_READ(13),
[CTRL] = e1000e_get_ctrl,
[TARC1] = e1000e_get_tarc,
[SWSM] = e1000e_mac_swsm_read,
@@ -3108,10 +3094,10 @@ static const readops e1000e_macreg_readops[] = {
[WUPM ... WUPM + 31] = e1000e_mac_readreg,
[MTA ... MTA + 127] = e1000e_mac_readreg,
[VFTA ... VFTA + 127] = e1000e_mac_readreg,
- [FFMT ... FFMT + 254] = E1000E_LOW_BITS_READ(4),
+ [FFMT ... FFMT + 254] = e1000e_mac_readreg,
[FFVT ... FFVT + 254] = e1000e_mac_readreg,
[MDEF ... MDEF + 7] = e1000e_mac_readreg,
- [FFLT ... FFLT + 10] = E1000E_LOW_BITS_READ(11),
+ [FFLT ... FFLT + 10] = e1000e_mac_readreg,
[FTFT ... FTFT + 254] = e1000e_mac_readreg,
[PBM ... PBM + 10239] = e1000e_mac_readreg,
[RETA ... RETA + 31] = e1000e_mac_readreg,
@@ -3134,19 +3120,8 @@ static const writeops e1000e_macreg_writeops[] = {
e1000e_putreg(LEDCTL),
e1000e_putreg(FCAL),
e1000e_putreg(FCRUC),
- e1000e_putreg(AIT),
- e1000e_putreg(TDFH),
- e1000e_putreg(TDFT),
- e1000e_putreg(TDFHS),
- e1000e_putreg(TDFTS),
- e1000e_putreg(TDFPC),
e1000e_putreg(WUC),
e1000e_putreg(WUS),
- e1000e_putreg(RDFH),
- e1000e_putreg(RDFT),
- e1000e_putreg(RDFHS),
- e1000e_putreg(RDFTS),
- e1000e_putreg(RDFPC),
e1000e_putreg(IPAV),
e1000e_putreg(TDBAH1),
e1000e_putreg(TIMINCA),
@@ -3157,7 +3132,6 @@ static const writeops e1000e_macreg_writeops[] = {
e1000e_putreg(TARC1),
e1000e_putreg(FLSWDATA),
e1000e_putreg(POEMB),
- e1000e_putreg(PBS),
e1000e_putreg(MFUTP01),
e1000e_putreg(MFUTP23),
e1000e_putreg(MANC),
@@ -3222,6 +3196,18 @@ static const writeops e1000e_macreg_writeops[] = {
[TADV] = e1000e_set_16bit,
[ITR] = e1000e_set_itr,
[EERD] = e1000e_set_eerd,
+ [AIT] = e1000e_set_16bit,
+ [TDFH] = e1000e_set_13bit,
+ [TDFT] = e1000e_set_13bit,
+ [TDFHS] = e1000e_set_13bit,
+ [TDFTS] = e1000e_set_13bit,
+ [TDFPC] = e1000e_set_13bit,
+ [RDFH] = e1000e_set_13bit,
+ [RDFHS] = e1000e_set_13bit,
+ [RDFT] = e1000e_set_13bit,
+ [RDFTS] = e1000e_set_13bit,
+ [RDFPC] = e1000e_set_13bit,
+ [PBS] = e1000e_set_6bit,
[GCR] = e1000e_set_gcr,
[PSRCTL] = e1000e_set_psrctl,
[RXCSUM] = e1000e_set_rxcsum,
@@ -3261,11 +3247,11 @@ static const writeops e1000e_macreg_writeops[] = {
[WUPM ... WUPM + 31] = e1000e_mac_writereg,
[MTA ... MTA + 127] = e1000e_mac_writereg,
[VFTA ... VFTA + 127] = e1000e_mac_writereg,
- [FFMT ... FFMT + 254] = e1000e_mac_writereg,
+ [FFMT ... FFMT + 254] = e1000e_set_4bit,
[FFVT ... FFVT + 254] = e1000e_mac_writereg,
[PBM ... PBM + 10239] = e1000e_mac_writereg,
[MDEF ... MDEF + 7] = e1000e_mac_writereg,
- [FFLT ... FFLT + 10] = e1000e_mac_writereg,
+ [FFLT ... FFLT + 10] = e1000e_set_11bit,
[FTFT ... FTFT + 254] = e1000e_mac_writereg,
[RETA ... RETA + 31] = e1000e_mac_writereg,
[RSSRK ... RSSRK + 31] = e1000e_mac_writereg,
--
2.39.1
^ permalink raw reply related [flat|nested] 33+ messages in thread* [PATCH v4 07/28] e1000: Use more constant definitions
2023-01-30 13:46 [PATCH v4 00/28] e1000x cleanups (preliminary for IGB) Akihiko Odaki
` (5 preceding siblings ...)
2023-01-30 13:46 ` [PATCH v4 06/28] e1000e: " Akihiko Odaki
@ 2023-01-30 13:46 ` Akihiko Odaki
2023-01-30 13:46 ` [PATCH v4 08/28] e1000e: " Akihiko Odaki
` (20 subsequent siblings)
27 siblings, 0 replies; 33+ messages in thread
From: Akihiko Odaki @ 2023-01-30 13:46 UTC (permalink / raw)
Cc: Jason Wang, Dmitry Fleytman, Thomas Huth, Laurent Vivier,
Paolo Bonzini, qemu-devel, qemu-ppc, Michael S . Tsirkin,
Yan Vugenfirer, Yuri Benditovich, Sriram Yagnaraman,
Alexander Bulekov, Akihiko Odaki, Philippe Mathieu-Daudé
The definitions for E1000_VFTA_ENTRY_SHIFT, E1000_VFTA_ENTRY_MASK, and
E1000_VFTA_ENTRY_BIT_SHIFT_MASK were copied from:
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/tree/drivers/net/ethernet/intel/e1000/e1000_hw.h?h=v6.0.9#n306
The definitions for E1000_NUM_UNICAST, E1000_MC_TBL_SIZE, and
E1000_VLAN_FILTER_TBL_SIZE were copied from:
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/tree/drivers/net/ethernet/intel/e1000/e1000_hw.h?h=v6.0.9#n707
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
---
hw/net/e1000.c | 50 +++++++++++++++++++++++-------------------
hw/net/e1000_regs.h | 9 ++++++++
hw/net/e1000x_common.c | 5 +++--
hw/net/e1000x_common.h | 2 +-
4 files changed, 41 insertions(+), 25 deletions(-)
diff --git a/hw/net/e1000.c b/hw/net/e1000.c
index 0925a99511..d9d048f665 100644
--- a/hw/net/e1000.c
+++ b/hw/net/e1000.c
@@ -43,8 +43,6 @@
#include "trace.h"
#include "qom/object.h"
-static const uint8_t bcast[] = {0xff, 0xff, 0xff, 0xff, 0xff, 0xff};
-
/* #define E1000_DEBUG */
#ifdef E1000_DEBUG
@@ -67,9 +65,8 @@ static int debugflags = DBGBIT(TXERR) | DBGBIT(GENERAL);
#define IOPORT_SIZE 0x40
#define PNPMMIO_SIZE 0x20000
-#define MIN_BUF_SIZE 60 /* Min. octets in an ethernet frame sans FCS */
-#define MAXIMUM_ETHERNET_HDR_LEN (14+4)
+#define MAXIMUM_ETHERNET_HDR_LEN (ETH_HLEN + 4)
/*
* HW models:
@@ -239,10 +236,16 @@ static const uint16_t phy_reg_init[] = {
[MII_PHYID1] = 0x141,
/* [MII_PHYID2] configured per DevId, from e1000_reset() */
- [MII_ANAR] = 0xde1,
- [MII_ANLPAR] = 0x1e0,
- [MII_CTRL1000] = 0x0e00,
- [MII_STAT1000] = 0x3c00,
+ [MII_ANAR] = MII_ANAR_CSMACD | MII_ANAR_10 |
+ MII_ANAR_10FD | MII_ANAR_TX |
+ MII_ANAR_TXFD | MII_ANAR_PAUSE |
+ MII_ANAR_PAUSE_ASYM,
+ [MII_ANLPAR] = MII_ANLPAR_10 | MII_ANLPAR_10FD |
+ MII_ANLPAR_TX | MII_ANLPAR_TXFD,
+ [MII_CTRL1000] = MII_CTRL1000_FULL | MII_CTRL1000_PORT |
+ MII_CTRL1000_MASTER,
+ [MII_STAT1000] = MII_STAT1000_HALF | MII_STAT1000_FULL |
+ MII_STAT1000_ROK | MII_STAT1000_LOK,
[M88E1000_PHY_SPEC_CTRL] = 0x360,
[M88E1000_PHY_SPEC_STATUS] = 0xac00,
[M88E1000_EXT_PHY_SPEC_CTRL] = 0x0d60,
@@ -548,9 +551,9 @@ putsum(uint8_t *data, uint32_t n, uint32_t sloc, uint32_t css, uint32_t cse)
static inline void
inc_tx_bcast_or_mcast_count(E1000State *s, const unsigned char *arr)
{
- if (!memcmp(arr, bcast, sizeof bcast)) {
+ if (is_broadcast_ether_addr(arr)) {
e1000x_inc_reg_if_not_full(s->mac_reg, BPTC);
- } else if (arr[0] & 1) {
+ } else if (is_multicast_ether_addr(arr)) {
e1000x_inc_reg_if_not_full(s->mac_reg, MPTC);
}
}
@@ -804,14 +807,16 @@ static int
receive_filter(E1000State *s, const uint8_t *buf, int size)
{
uint32_t rctl = s->mac_reg[RCTL];
- int isbcast = !memcmp(buf, bcast, sizeof bcast), ismcast = (buf[0] & 1);
+ int isbcast = is_broadcast_ether_addr(buf);
+ int ismcast = is_multicast_ether_addr(buf);
if (e1000x_is_vlan_packet(buf, le16_to_cpu(s->mac_reg[VET])) &&
e1000x_vlan_rx_filter_enabled(s->mac_reg)) {
- uint16_t vid = lduw_be_p(buf + 14);
- uint32_t vfta = ldl_le_p((uint32_t *)(s->mac_reg + VFTA) +
- ((vid >> 5) & 0x7f));
- if ((vfta & (1 << (vid & 0x1f))) == 0) {
+ uint16_t vid = lduw_be_p(&PKT_GET_VLAN_HDR(buf)->h_tci);
+ uint32_t vfta =
+ ldl_le_p((uint32_t *)(s->mac_reg + VFTA) +
+ ((vid >> E1000_VFTA_ENTRY_SHIFT) & E1000_VFTA_ENTRY_MASK));
+ if ((vfta & (1 << (vid & E1000_VFTA_ENTRY_BIT_SHIFT_MASK))) == 0) {
return 0;
}
}
@@ -909,7 +914,7 @@ e1000_receive_iov(NetClientState *nc, const struct iovec *iov, int iovcnt)
uint32_t rdh_start;
uint16_t vlan_special = 0;
uint8_t vlan_status = 0;
- uint8_t min_buf[MIN_BUF_SIZE];
+ uint8_t min_buf[ETH_ZLEN];
struct iovec min_iov;
uint8_t *filter_buf = iov->iov_base;
size_t size = iov_size(iov, iovcnt);
@@ -1204,8 +1209,8 @@ static const readops macreg_readops[] = {
[FFLT ... FFLT + 6] = &mac_readreg,
[RA ... RA + 31] = &mac_readreg,
[WUPM ... WUPM + 31] = &mac_readreg,
- [MTA ... MTA + 127] = &mac_readreg,
- [VFTA ... VFTA + 127] = &mac_readreg,
+ [MTA ... MTA + E1000_MC_TBL_SIZE - 1] = &mac_readreg,
+ [VFTA ... VFTA + E1000_VLAN_FILTER_TBL_SIZE - 1] = &mac_readreg,
[FFMT ... FFMT + 254] = &mac_readreg,
[FFVT ... FFVT + 254] = &mac_readreg,
[PBM ... PBM + 16383] = &mac_readreg,
@@ -1236,8 +1241,8 @@ static const writeops macreg_writeops[] = {
[FFLT ... FFLT + 6] = &set_11bit,
[RA ... RA + 31] = &mac_writereg,
[WUPM ... WUPM + 31] = &mac_writereg,
- [MTA ... MTA + 127] = &mac_writereg,
- [VFTA ... VFTA + 127] = &mac_writereg,
+ [MTA ... MTA + E1000_MC_TBL_SIZE - 1] = &mac_writereg,
+ [VFTA ... VFTA + E1000_VLAN_FILTER_TBL_SIZE - 1] = &mac_writereg,
[FFMT ... FFMT + 254] = &set_4bit, [FFVT ... FFVT + 254] = &mac_writereg,
[PBM ... PBM + 16383] = &mac_writereg,
};
@@ -1603,8 +1608,9 @@ static const VMStateDescription vmstate_e1000 = {
VMSTATE_UINT32(mac_reg[WUFC], E1000State),
VMSTATE_UINT32(mac_reg[VET], E1000State),
VMSTATE_UINT32_SUB_ARRAY(mac_reg, E1000State, RA, 32),
- VMSTATE_UINT32_SUB_ARRAY(mac_reg, E1000State, MTA, 128),
- VMSTATE_UINT32_SUB_ARRAY(mac_reg, E1000State, VFTA, 128),
+ VMSTATE_UINT32_SUB_ARRAY(mac_reg, E1000State, MTA, E1000_MC_TBL_SIZE),
+ VMSTATE_UINT32_SUB_ARRAY(mac_reg, E1000State, VFTA,
+ E1000_VLAN_FILTER_TBL_SIZE),
VMSTATE_END_OF_LIST()
},
.subsections = (const VMStateDescription*[]) {
diff --git a/hw/net/e1000_regs.h b/hw/net/e1000_regs.h
index 59d6eb3189..3f6b5d0c52 100644
--- a/hw/net/e1000_regs.h
+++ b/hw/net/e1000_regs.h
@@ -1154,6 +1154,11 @@ struct e1000_data_desc {
} upper;
};
+/* Filters */
+#define E1000_NUM_UNICAST 16 /* Unicast filter entries */
+#define E1000_MC_TBL_SIZE 128 /* Multicast Filter Table (4096 bits) */
+#define E1000_VLAN_FILTER_TBL_SIZE 128 /* VLAN Filter Table (4096 bits) */
+
/* Management Control */
#define E1000_MANC_SMBUS_EN 0x00000001 /* SMBus Enabled - RO */
#define E1000_MANC_ASF_EN 0x00000002 /* ASF Enabled - RO */
@@ -1202,4 +1207,8 @@ struct e1000_data_desc {
#define E1000_IOADDR 0x00
#define E1000_IODATA 0x04
+#define E1000_VFTA_ENTRY_SHIFT 5
+#define E1000_VFTA_ENTRY_MASK 0x7F
+#define E1000_VFTA_ENTRY_BIT_SHIFT_MASK 0x1F
+
#endif /* HW_E1000_REGS_H */
diff --git a/hw/net/e1000x_common.c b/hw/net/e1000x_common.c
index e79d4c79bd..b3bbf31582 100644
--- a/hw/net/e1000x_common.c
+++ b/hw/net/e1000x_common.c
@@ -26,6 +26,7 @@
#include "qemu/units.h"
#include "hw/net/mii.h"
#include "hw/pci/pci_device.h"
+#include "net/eth.h"
#include "net/net.h"
#include "e1000x_common.h"
@@ -48,7 +49,7 @@ bool e1000x_rx_ready(PCIDevice *d, uint32_t *mac)
bool e1000x_is_vlan_packet(const uint8_t *buf, uint16_t vet)
{
- uint16_t eth_proto = lduw_be_p(buf + 12);
+ uint16_t eth_proto = lduw_be_p(&PKT_GET_ETH_HDR(buf)->h_proto);
bool res = (eth_proto == vet);
trace_e1000x_vlan_is_vlan_pkt(res, eth_proto, vet);
@@ -67,7 +68,7 @@ bool e1000x_rx_group_filter(uint32_t *mac, const uint8_t *buf)
}
ra[0] = cpu_to_le32(rp[0]);
ra[1] = cpu_to_le32(rp[1]);
- if (!memcmp(buf, (uint8_t *)ra, 6)) {
+ if (!memcmp(buf, (uint8_t *)ra, ETH_ALEN)) {
trace_e1000x_rx_flt_ucast_match((int)(rp - mac - RA) / 2,
MAC_ARG(buf));
return true;
diff --git a/hw/net/e1000x_common.h b/hw/net/e1000x_common.h
index 3501e4855a..b991d814b1 100644
--- a/hw/net/e1000x_common.h
+++ b/hw/net/e1000x_common.h
@@ -102,7 +102,7 @@ enum {
static inline void
e1000x_inc_reg_if_not_full(uint32_t *mac, int index)
{
- if (mac[index] != 0xffffffff) {
+ if (mac[index] != UINT32_MAX) {
mac[index]++;
}
}
--
2.39.1
^ permalink raw reply related [flat|nested] 33+ messages in thread* [PATCH v4 08/28] e1000e: Use more constant definitions
2023-01-30 13:46 [PATCH v4 00/28] e1000x cleanups (preliminary for IGB) Akihiko Odaki
` (6 preceding siblings ...)
2023-01-30 13:46 ` [PATCH v4 07/28] e1000: Use more constant definitions Akihiko Odaki
@ 2023-01-30 13:46 ` Akihiko Odaki
2023-01-30 13:46 ` [PATCH v4 09/28] e1000: Use memcpy to intialize registers Akihiko Odaki
` (19 subsequent siblings)
27 siblings, 0 replies; 33+ messages in thread
From: Akihiko Odaki @ 2023-01-30 13:46 UTC (permalink / raw)
Cc: Jason Wang, Dmitry Fleytman, Thomas Huth, Laurent Vivier,
Paolo Bonzini, qemu-devel, qemu-ppc, Michael S . Tsirkin,
Yan Vugenfirer, Yuri Benditovich, Sriram Yagnaraman,
Alexander Bulekov, Akihiko Odaki
The definitions of SW Semaphore Register were copied from:
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/tree/drivers/net/ethernet/intel/e1000e/defines.h?h=v6.0.9#n374
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
---
hw/net/e1000_regs.h | 7 +++++++
hw/net/e1000e_core.c | 49 ++++++++++++++++++++++++--------------------
2 files changed, 34 insertions(+), 22 deletions(-)
diff --git a/hw/net/e1000_regs.h b/hw/net/e1000_regs.h
index 3f6b5d0c52..6a36573802 100644
--- a/hw/net/e1000_regs.h
+++ b/hw/net/e1000_regs.h
@@ -525,6 +525,13 @@
#define M88E1000_PHY_VCO_REG_BIT8 0x100 /* Bits 8 & 11 are adjusted for */
#define M88E1000_PHY_VCO_REG_BIT11 0x800 /* improved BER performance */
+/* SW Semaphore Register */
+#define E1000_SWSM_SMBI 0x00000001 /* Driver Semaphore bit */
+#define E1000_SWSM_SWESMBI 0x00000002 /* FW Semaphore bit */
+#define E1000_SWSM_DRV_LOAD 0x00000008 /* Driver Loaded Bit */
+
+#define E1000_SWSM2_LOCK 0x00000002 /* Secondary driver semaphore bit */
+
/* Interrupt Cause Read */
#define E1000_ICR_TXDW 0x00000001 /* Transmit desc written back */
#define E1000_ICR_TXQE 0x00000002 /* Transmit Queue empty */
diff --git a/hw/net/e1000e_core.c b/hw/net/e1000e_core.c
index e6fc85ea51..6a4da72bd3 100644
--- a/hw/net/e1000e_core.c
+++ b/hw/net/e1000e_core.c
@@ -1022,10 +1022,11 @@ e1000e_receive_filter(E1000ECore *core, const uint8_t *buf, int size)
if (e1000x_is_vlan_packet(buf, core->mac[VET]) &&
e1000x_vlan_rx_filter_enabled(core->mac)) {
- uint16_t vid = lduw_be_p(buf + 14);
- uint32_t vfta = ldl_le_p((uint32_t *)(core->mac + VFTA) +
- ((vid >> 5) & 0x7f));
- if ((vfta & (1 << (vid & 0x1f))) == 0) {
+ uint16_t vid = lduw_be_p(&PKT_GET_VLAN_HDR(buf)->h_tci);
+ uint32_t vfta =
+ ldl_le_p((uint32_t *)(core->mac + VFTA) +
+ ((vid >> E1000_VFTA_ENTRY_SHIFT) & E1000_VFTA_ENTRY_MASK));
+ if ((vfta & (1 << (vid & E1000_VFTA_ENTRY_BIT_SHIFT_MASK))) == 0) {
trace_e1000e_rx_flt_vlan_mismatch(vid);
return false;
} else {
@@ -1679,16 +1680,13 @@ e1000e_rx_fix_l4_csum(E1000ECore *core, struct NetRxPkt *pkt)
}
}
-/* Min. octets in an ethernet frame sans FCS */
-#define MIN_BUF_SIZE 60
-
ssize_t
e1000e_receive_iov(E1000ECore *core, const struct iovec *iov, int iovcnt)
{
- static const int maximum_ethernet_hdr_len = (14 + 4);
+ static const int maximum_ethernet_hdr_len = (ETH_HLEN + 4);
uint32_t n = 0;
- uint8_t min_buf[MIN_BUF_SIZE];
+ uint8_t min_buf[ETH_ZLEN];
struct iovec min_iov;
uint8_t *filter_buf;
size_t size, orig_size;
@@ -2627,7 +2625,7 @@ static uint32_t
e1000e_mac_swsm_read(E1000ECore *core, int index)
{
uint32_t val = core->mac[SWSM];
- core->mac[SWSM] = val | 1;
+ core->mac[SWSM] = val | E1000_SWSM_SMBI;
return val;
}
@@ -3092,8 +3090,8 @@ static const readops e1000e_macreg_readops[] = {
[IP4AT ... IP4AT + 6] = e1000e_mac_readreg,
[RA ... RA + 31] = e1000e_mac_readreg,
[WUPM ... WUPM + 31] = e1000e_mac_readreg,
- [MTA ... MTA + 127] = e1000e_mac_readreg,
- [VFTA ... VFTA + 127] = e1000e_mac_readreg,
+ [MTA ... MTA + E1000_MC_TBL_SIZE - 1] = e1000e_mac_readreg,
+ [VFTA ... VFTA + E1000_VLAN_FILTER_TBL_SIZE - 1] = e1000e_mac_readreg,
[FFMT ... FFMT + 254] = e1000e_mac_readreg,
[FFVT ... FFVT + 254] = e1000e_mac_readreg,
[MDEF ... MDEF + 7] = e1000e_mac_readreg,
@@ -3245,8 +3243,8 @@ static const writeops e1000e_macreg_writeops[] = {
[IP4AT ... IP4AT + 6] = e1000e_mac_writereg,
[RA + 2 ... RA + 31] = e1000e_mac_writereg,
[WUPM ... WUPM + 31] = e1000e_mac_writereg,
- [MTA ... MTA + 127] = e1000e_mac_writereg,
- [VFTA ... VFTA + 127] = e1000e_mac_writereg,
+ [MTA ... MTA + E1000_MC_TBL_SIZE - 1] = e1000e_mac_writereg,
+ [VFTA ... VFTA + E1000_VLAN_FILTER_TBL_SIZE - 1] = e1000e_mac_writereg,
[FFMT ... FFMT + 254] = e1000e_set_4bit,
[FFVT ... FFVT + 254] = e1000e_mac_writereg,
[PBM ... PBM + 10239] = e1000e_mac_writereg,
@@ -3276,7 +3274,7 @@ static const uint16_t mac_reg_access[E1000E_MAC_SIZE] = {
[TDH_A] = 0x0cf8, [TDT_A] = 0x0cf8, [TIDV_A] = 0x0cf8,
[TDFH_A] = 0xed00, [TDFT_A] = 0xed00,
[RA_A ... RA_A + 31] = 0x14f0,
- [VFTA_A ... VFTA_A + 127] = 0x1400,
+ [VFTA_A ... VFTA_A + E1000_VLAN_FILTER_TBL_SIZE - 1] = 0x1400,
[RDBAL0_A ... RDLEN0_A] = 0x09bc,
[TDBAL_A ... TDLEN_A] = 0x0cf8,
/* Access options */
@@ -3433,13 +3431,20 @@ e1000e_phy_reg_init[E1000E_PHY_PAGES][E1000E_PHY_PAGE_SIZE] = {
[MII_PHYID1] = 0x141,
[MII_PHYID2] = E1000_PHY_ID2_82574x,
- [MII_ANAR] = 0xde1,
- [MII_ANLPAR] = 0x7e0,
- [MII_ANER] = BIT(2),
- [MII_ANNP] = BIT(0) | BIT(13),
- [MII_CTRL1000] = BIT(8) | BIT(9) | BIT(10) | BIT(11),
- [MII_STAT1000] = 0x3c00,
- [MII_EXTSTAT] = BIT(12) | BIT(13),
+ [MII_ANAR] = MII_ANAR_CSMACD | MII_ANAR_10 |
+ MII_ANAR_10FD | MII_ANAR_TX |
+ MII_ANAR_TXFD | MII_ANAR_PAUSE |
+ MII_ANAR_PAUSE_ASYM,
+ [MII_ANLPAR] = MII_ANLPAR_10 | MII_ANLPAR_10FD |
+ MII_ANLPAR_TX | MII_ANLPAR_TXFD |
+ MII_ANLPAR_T4 | MII_ANLPAR_PAUSE,
+ [MII_ANER] = MII_ANER_NP,
+ [MII_ANNP] = 1 | MII_ANNP_MP,
+ [MII_CTRL1000] = MII_CTRL1000_HALF | MII_CTRL1000_FULL |
+ MII_CTRL1000_PORT | MII_CTRL1000_MASTER,
+ [MII_STAT1000] = MII_STAT1000_HALF | MII_STAT1000_FULL |
+ MII_STAT1000_ROK | MII_STAT1000_LOK,
+ [MII_EXTSTAT] = MII_EXTSTAT_1000T_HD | MII_EXTSTAT_1000T_FD,
[PHY_COPPER_CTRL1] = BIT(5) | BIT(6) | BIT(8) | BIT(9) |
BIT(12) | BIT(13),
--
2.39.1
^ permalink raw reply related [flat|nested] 33+ messages in thread* [PATCH v4 09/28] e1000: Use memcpy to intialize registers
2023-01-30 13:46 [PATCH v4 00/28] e1000x cleanups (preliminary for IGB) Akihiko Odaki
` (7 preceding siblings ...)
2023-01-30 13:46 ` [PATCH v4 08/28] e1000e: " Akihiko Odaki
@ 2023-01-30 13:46 ` Akihiko Odaki
2023-01-30 13:46 ` [PATCH v4 10/28] e1000e: " Akihiko Odaki
` (18 subsequent siblings)
27 siblings, 0 replies; 33+ messages in thread
From: Akihiko Odaki @ 2023-01-30 13:46 UTC (permalink / raw)
Cc: Jason Wang, Dmitry Fleytman, Thomas Huth, Laurent Vivier,
Paolo Bonzini, qemu-devel, qemu-ppc, Michael S . Tsirkin,
Yan Vugenfirer, Yuri Benditovich, Sriram Yagnaraman,
Alexander Bulekov, Akihiko Odaki
Use memcpy instead of memmove to initialize registers. The initial
register templates and register table instances will never overlap.
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
---
hw/net/e1000.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/hw/net/e1000.c b/hw/net/e1000.c
index d9d048f665..3353a3752c 100644
--- a/hw/net/e1000.c
+++ b/hw/net/e1000.c
@@ -390,10 +390,10 @@ static void e1000_reset(void *opaque)
d->mit_irq_level = 0;
d->mit_ide = 0;
memset(d->phy_reg, 0, sizeof d->phy_reg);
- memmove(d->phy_reg, phy_reg_init, sizeof phy_reg_init);
+ memcpy(d->phy_reg, phy_reg_init, sizeof phy_reg_init);
d->phy_reg[MII_PHYID2] = edc->phy_id2;
memset(d->mac_reg, 0, sizeof d->mac_reg);
- memmove(d->mac_reg, mac_reg_init, sizeof mac_reg_init);
+ memcpy(d->mac_reg, mac_reg_init, sizeof mac_reg_init);
d->rxbuf_min_shift = 1;
memset(&d->tx, 0, sizeof d->tx);
--
2.39.1
^ permalink raw reply related [flat|nested] 33+ messages in thread* [PATCH v4 10/28] e1000e: Use memcpy to intialize registers
2023-01-30 13:46 [PATCH v4 00/28] e1000x cleanups (preliminary for IGB) Akihiko Odaki
` (8 preceding siblings ...)
2023-01-30 13:46 ` [PATCH v4 09/28] e1000: Use memcpy to intialize registers Akihiko Odaki
@ 2023-01-30 13:46 ` Akihiko Odaki
2023-01-30 13:46 ` [PATCH v4 11/28] e1000e: Remove pending interrupt flags Akihiko Odaki
` (17 subsequent siblings)
27 siblings, 0 replies; 33+ messages in thread
From: Akihiko Odaki @ 2023-01-30 13:46 UTC (permalink / raw)
Cc: Jason Wang, Dmitry Fleytman, Thomas Huth, Laurent Vivier,
Paolo Bonzini, qemu-devel, qemu-ppc, Michael S . Tsirkin,
Yan Vugenfirer, Yuri Benditovich, Sriram Yagnaraman,
Alexander Bulekov, Akihiko Odaki
Use memcpy instead of memmove to initialize registers. The initial
register templates and register table instances will never overlap.
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
---
hw/net/e1000e_core.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/hw/net/e1000e_core.c b/hw/net/e1000e_core.c
index 6a4da72bd3..87f964cdc1 100644
--- a/hw/net/e1000e_core.c
+++ b/hw/net/e1000e_core.c
@@ -3511,9 +3511,9 @@ e1000e_core_reset(E1000ECore *core)
e1000e_intrmgr_reset(core);
memset(core->phy, 0, sizeof core->phy);
- memmove(core->phy, e1000e_phy_reg_init, sizeof e1000e_phy_reg_init);
+ memcpy(core->phy, e1000e_phy_reg_init, sizeof e1000e_phy_reg_init);
memset(core->mac, 0, sizeof core->mac);
- memmove(core->mac, e1000e_mac_reg_init, sizeof e1000e_mac_reg_init);
+ memcpy(core->mac, e1000e_mac_reg_init, sizeof e1000e_mac_reg_init);
core->rxbuf_min_shift = 1 + E1000_RING_DESC_LEN_SHIFT;
--
2.39.1
^ permalink raw reply related [flat|nested] 33+ messages in thread* [PATCH v4 11/28] e1000e: Remove pending interrupt flags
2023-01-30 13:46 [PATCH v4 00/28] e1000x cleanups (preliminary for IGB) Akihiko Odaki
` (9 preceding siblings ...)
2023-01-30 13:46 ` [PATCH v4 10/28] e1000e: " Akihiko Odaki
@ 2023-01-30 13:46 ` Akihiko Odaki
2023-01-30 13:46 ` [PATCH v4 12/28] e1000e: Improve software reset Akihiko Odaki
` (16 subsequent siblings)
27 siblings, 0 replies; 33+ messages in thread
From: Akihiko Odaki @ 2023-01-30 13:46 UTC (permalink / raw)
Cc: Jason Wang, Dmitry Fleytman, Thomas Huth, Laurent Vivier,
Paolo Bonzini, qemu-devel, qemu-ppc, Michael S . Tsirkin,
Yan Vugenfirer, Yuri Benditovich, Sriram Yagnaraman,
Alexander Bulekov, Akihiko Odaki
They are duplicate of running throttling timer flags and incomplete as
the flags are not cleared when the interrupts are fired or the device is
reset.
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
---
hw/net/e1000e.c | 5 ++---
hw/net/e1000e_core.c | 19 +++----------------
hw/net/e1000e_core.h | 2 --
hw/net/trace-events | 2 --
4 files changed, 5 insertions(+), 23 deletions(-)
diff --git a/hw/net/e1000e.c b/hw/net/e1000e.c
index d591d01c07..0bc222d354 100644
--- a/hw/net/e1000e.c
+++ b/hw/net/e1000e.c
@@ -631,12 +631,11 @@ static const VMStateDescription e1000e_vmstate = {
VMSTATE_E1000E_INTR_DELAY_TIMER(core.tidv, E1000EState),
VMSTATE_E1000E_INTR_DELAY_TIMER(core.itr, E1000EState),
- VMSTATE_BOOL(core.itr_intr_pending, E1000EState),
+ VMSTATE_UNUSED(1),
VMSTATE_E1000E_INTR_DELAY_TIMER_ARRAY(core.eitr, E1000EState,
E1000E_MSIX_VEC_NUM),
- VMSTATE_BOOL_ARRAY(core.eitr_intr_pending, E1000EState,
- E1000E_MSIX_VEC_NUM),
+ VMSTATE_UNUSED(E1000E_MSIX_VEC_NUM),
VMSTATE_UINT32(core.itr_guest_value, E1000EState),
VMSTATE_UINT32_ARRAY(core.eitr_guest_value, E1000EState,
diff --git a/hw/net/e1000e_core.c b/hw/net/e1000e_core.c
index 87f964cdc1..37aec6a970 100644
--- a/hw/net/e1000e_core.c
+++ b/hw/net/e1000e_core.c
@@ -154,11 +154,6 @@ e1000e_intrmgr_on_throttling_timer(void *opaque)
timer->running = false;
- if (!timer->core->itr_intr_pending) {
- trace_e1000e_irq_throttling_no_pending_interrupts();
- return;
- }
-
if (msi_enabled(timer->core->owner)) {
trace_e1000e_irq_msi_notify_postponed();
/* Clear msi_causes_pending to fire MSI eventually */
@@ -180,11 +175,6 @@ e1000e_intrmgr_on_msix_throttling_timer(void *opaque)
timer->running = false;
- if (!timer->core->eitr_intr_pending[idx]) {
- trace_e1000e_irq_throttling_no_pending_vec(idx);
- return;
- }
-
trace_e1000e_irq_msix_notify_postponed_vec(idx);
msix_notify(timer->core->owner, idx);
}
@@ -2015,13 +2005,11 @@ e1000e_clear_ims_bits(E1000ECore *core, uint32_t bits)
}
static inline bool
-e1000e_postpone_interrupt(bool *interrupt_pending,
- E1000IntrDelayTimer *timer)
+e1000e_postpone_interrupt(E1000IntrDelayTimer *timer)
{
if (timer->running) {
trace_e1000e_irq_postponed_by_xitr(timer->delay_reg << 2);
- *interrupt_pending = true;
return true;
}
@@ -2035,14 +2023,13 @@ e1000e_postpone_interrupt(bool *interrupt_pending,
static inline bool
e1000e_itr_should_postpone(E1000ECore *core)
{
- return e1000e_postpone_interrupt(&core->itr_intr_pending, &core->itr);
+ return e1000e_postpone_interrupt(&core->itr);
}
static inline bool
e1000e_eitr_should_postpone(E1000ECore *core, int idx)
{
- return e1000e_postpone_interrupt(&core->eitr_intr_pending[idx],
- &core->eitr[idx]);
+ return e1000e_postpone_interrupt(&core->eitr[idx]);
}
static void
diff --git a/hw/net/e1000e_core.h b/hw/net/e1000e_core.h
index b8f38c47a0..d0a14b4523 100644
--- a/hw/net/e1000e_core.h
+++ b/hw/net/e1000e_core.h
@@ -95,10 +95,8 @@ struct E1000Core {
E1000IntrDelayTimer tidv;
E1000IntrDelayTimer itr;
- bool itr_intr_pending;
E1000IntrDelayTimer eitr[E1000E_MSIX_VEC_NUM];
- bool eitr_intr_pending[E1000E_MSIX_VEC_NUM];
VMChangeStateEntry *vmstate;
diff --git a/hw/net/trace-events b/hw/net/trace-events
index 4c0ec3fda1..8fa4299704 100644
--- a/hw/net/trace-events
+++ b/hw/net/trace-events
@@ -201,10 +201,8 @@ e1000e_rx_metadata_ipv6_filtering_disabled(void) "IPv6 RX filtering disabled by
e1000e_vlan_vet(uint16_t vet) "Setting VLAN ethernet type 0x%X"
e1000e_irq_msi_notify(uint32_t cause) "MSI notify 0x%x"
-e1000e_irq_throttling_no_pending_interrupts(void) "No pending interrupts to notify"
e1000e_irq_msi_notify_postponed(void) "Sending MSI postponed by ITR"
e1000e_irq_legacy_notify_postponed(void) "Raising legacy IRQ postponed by ITR"
-e1000e_irq_throttling_no_pending_vec(int idx) "No pending interrupts for vector %d"
e1000e_irq_msix_notify_postponed_vec(int idx) "Sending MSI-X postponed by EITR[%d]"
e1000e_irq_legacy_notify(bool level) "IRQ line state: %d"
e1000e_irq_msix_notify_vec(uint32_t vector) "MSI-X notify vector 0x%x"
--
2.39.1
^ permalink raw reply related [flat|nested] 33+ messages in thread* [PATCH v4 12/28] e1000e: Improve software reset
2023-01-30 13:46 [PATCH v4 00/28] e1000x cleanups (preliminary for IGB) Akihiko Odaki
` (10 preceding siblings ...)
2023-01-30 13:46 ` [PATCH v4 11/28] e1000e: Remove pending interrupt flags Akihiko Odaki
@ 2023-01-30 13:46 ` Akihiko Odaki
2023-01-30 13:47 ` [PATCH v4 13/28] e1000: Configure ResettableClass Akihiko Odaki
` (15 subsequent siblings)
27 siblings, 0 replies; 33+ messages in thread
From: Akihiko Odaki @ 2023-01-30 13:46 UTC (permalink / raw)
Cc: Jason Wang, Dmitry Fleytman, Thomas Huth, Laurent Vivier,
Paolo Bonzini, qemu-devel, qemu-ppc, Michael S . Tsirkin,
Yan Vugenfirer, Yuri Benditovich, Sriram Yagnaraman,
Alexander Bulekov, Akihiko Odaki
This change makes e1000e reset more things when software reset was
triggered. Some registers are exempted from software reset in the
datasheet and this change also implements the behavior accordingly.
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
---
hw/net/e1000e_core.c | 24 +++++++++++++++++++-----
1 file changed, 19 insertions(+), 5 deletions(-)
diff --git a/hw/net/e1000e_core.c b/hw/net/e1000e_core.c
index 37aec6a970..b8670662c8 100644
--- a/hw/net/e1000e_core.c
+++ b/hw/net/e1000e_core.c
@@ -58,6 +58,8 @@
static inline void
e1000e_set_interrupt_cause(E1000ECore *core, uint32_t val);
+static void e1000e_reset(E1000ECore *core, bool sw);
+
static inline void
e1000e_process_ts_option(E1000ECore *core, struct e1000_tx_desc *dp)
{
@@ -1882,7 +1884,7 @@ e1000e_set_ctrl(E1000ECore *core, int index, uint32_t val)
if (val & E1000_CTRL_RST) {
trace_e1000e_core_ctrl_sw_reset();
- e1000x_reset_mac_addr(core->owner_nic, core->mac, core->permanent_mac);
+ e1000e_reset(core, true);
}
if (val & E1000_CTRL_PHY_RST) {
@@ -3488,8 +3490,7 @@ static const uint32_t e1000e_mac_reg_init[] = {
[EITR...EITR + E1000E_MSIX_VEC_NUM - 1] = E1000E_MIN_XITR,
};
-void
-e1000e_core_reset(E1000ECore *core)
+static void e1000e_reset(E1000ECore *core, bool sw)
{
int i;
@@ -3499,8 +3500,15 @@ e1000e_core_reset(E1000ECore *core)
memset(core->phy, 0, sizeof core->phy);
memcpy(core->phy, e1000e_phy_reg_init, sizeof e1000e_phy_reg_init);
- memset(core->mac, 0, sizeof core->mac);
- memcpy(core->mac, e1000e_mac_reg_init, sizeof e1000e_mac_reg_init);
+
+ for (i = 0; i < E1000E_MAC_SIZE; i++) {
+ if (sw && (i == PBA || i == PBS || i == FLA)) {
+ continue;
+ }
+
+ core->mac[i] = i < ARRAY_SIZE(e1000e_mac_reg_init) ?
+ e1000e_mac_reg_init[i] : 0;
+ }
core->rxbuf_min_shift = 1 + E1000_RING_DESC_LEN_SHIFT;
@@ -3517,6 +3525,12 @@ e1000e_core_reset(E1000ECore *core)
}
}
+void
+e1000e_core_reset(E1000ECore *core)
+{
+ e1000e_reset(core, false);
+}
+
void e1000e_core_pre_save(E1000ECore *core)
{
int i;
--
2.39.1
^ permalink raw reply related [flat|nested] 33+ messages in thread* [PATCH v4 13/28] e1000: Configure ResettableClass
2023-01-30 13:46 [PATCH v4 00/28] e1000x cleanups (preliminary for IGB) Akihiko Odaki
` (11 preceding siblings ...)
2023-01-30 13:46 ` [PATCH v4 12/28] e1000e: Improve software reset Akihiko Odaki
@ 2023-01-30 13:47 ` Akihiko Odaki
2023-01-30 13:47 ` [PATCH v4 14/28] e1000e: " Akihiko Odaki
` (14 subsequent siblings)
27 siblings, 0 replies; 33+ messages in thread
From: Akihiko Odaki @ 2023-01-30 13:47 UTC (permalink / raw)
Cc: Jason Wang, Dmitry Fleytman, Thomas Huth, Laurent Vivier,
Paolo Bonzini, qemu-devel, qemu-ppc, Michael S . Tsirkin,
Yan Vugenfirer, Yuri Benditovich, Sriram Yagnaraman,
Alexander Bulekov, Akihiko Odaki, Peter Maydell,
Philippe Mathieu-Daudé
This is part of recent efforts of refactoring e1000 and e1000e.
DeviceClass's reset member is deprecated so migrate to ResettableClass.
There is no behavioral difference.
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
---
hw/net/e1000.c | 13 ++++---------
1 file changed, 4 insertions(+), 9 deletions(-)
diff --git a/hw/net/e1000.c b/hw/net/e1000.c
index 3353a3752c..c81d914a02 100644
--- a/hw/net/e1000.c
+++ b/hw/net/e1000.c
@@ -377,9 +377,9 @@ static bool e1000_vet_init_need(void *opaque)
return chkflag(VET);
}
-static void e1000_reset(void *opaque)
+static void e1000_reset_hold(Object *obj)
{
- E1000State *d = opaque;
+ E1000State *d = E1000(obj);
E1000BaseClass *edc = E1000_GET_CLASS(d);
uint8_t *macaddr = d->conf.macaddr.a;
@@ -1731,12 +1731,6 @@ static void pci_e1000_realize(PCIDevice *pci_dev, Error **errp)
e1000_flush_queue_timer, d);
}
-static void qdev_e1000_reset(DeviceState *dev)
-{
- E1000State *d = E1000(dev);
- e1000_reset(d);
-}
-
static Property e1000_properties[] = {
DEFINE_NIC_PROPERTIES(E1000State, conf),
DEFINE_PROP_BIT("autonegotiation", E1000State,
@@ -1762,6 +1756,7 @@ typedef struct E1000Info {
static void e1000_class_init(ObjectClass *klass, void *data)
{
DeviceClass *dc = DEVICE_CLASS(klass);
+ ResettableClass *rc = RESETTABLE_CLASS(klass);
PCIDeviceClass *k = PCI_DEVICE_CLASS(klass);
E1000BaseClass *e = E1000_CLASS(klass);
const E1000Info *info = data;
@@ -1774,9 +1769,9 @@ static void e1000_class_init(ObjectClass *klass, void *data)
k->revision = info->revision;
e->phy_id2 = info->phy_id2;
k->class_id = PCI_CLASS_NETWORK_ETHERNET;
+ rc->phases.hold = e1000_reset_hold;
set_bit(DEVICE_CATEGORY_NETWORK, dc->categories);
dc->desc = "Intel Gigabit Ethernet";
- dc->reset = qdev_e1000_reset;
dc->vmsd = &vmstate_e1000;
device_class_set_props(dc, e1000_properties);
}
--
2.39.1
^ permalink raw reply related [flat|nested] 33+ messages in thread* [PATCH v4 14/28] e1000e: Configure ResettableClass
2023-01-30 13:46 [PATCH v4 00/28] e1000x cleanups (preliminary for IGB) Akihiko Odaki
` (12 preceding siblings ...)
2023-01-30 13:47 ` [PATCH v4 13/28] e1000: Configure ResettableClass Akihiko Odaki
@ 2023-01-30 13:47 ` Akihiko Odaki
2023-01-30 13:47 ` [PATCH v4 15/28] e1000e: Introduce e1000_rx_desc_union Akihiko Odaki
` (13 subsequent siblings)
27 siblings, 0 replies; 33+ messages in thread
From: Akihiko Odaki @ 2023-01-30 13:47 UTC (permalink / raw)
Cc: Jason Wang, Dmitry Fleytman, Thomas Huth, Laurent Vivier,
Paolo Bonzini, qemu-devel, qemu-ppc, Michael S . Tsirkin,
Yan Vugenfirer, Yuri Benditovich, Sriram Yagnaraman,
Alexander Bulekov, Akihiko Odaki, Peter Maydell,
Philippe Mathieu-Daudé
This is part of recent efforts of refactoring e1000 and e1000e.
DeviceClass's reset member is deprecated so migrate to ResettableClass.
There is no behavioral difference.
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
---
hw/net/e1000e.c | 10 ++++++----
hw/net/trace-events | 2 +-
2 files changed, 7 insertions(+), 5 deletions(-)
diff --git a/hw/net/e1000e.c b/hw/net/e1000e.c
index 0bc222d354..ec274319c4 100644
--- a/hw/net/e1000e.c
+++ b/hw/net/e1000e.c
@@ -513,11 +513,11 @@ static void e1000e_pci_uninit(PCIDevice *pci_dev)
msi_uninit(pci_dev);
}
-static void e1000e_qdev_reset(DeviceState *dev)
+static void e1000e_qdev_reset_hold(Object *obj)
{
- E1000EState *s = E1000E(dev);
+ E1000EState *s = E1000E(obj);
- trace_e1000e_cb_qdev_reset();
+ trace_e1000e_cb_qdev_reset_hold();
e1000e_core_reset(&s->core);
@@ -669,6 +669,7 @@ static Property e1000e_properties[] = {
static void e1000e_class_init(ObjectClass *class, void *data)
{
DeviceClass *dc = DEVICE_CLASS(class);
+ ResettableClass *rc = RESETTABLE_CLASS(class);
PCIDeviceClass *c = PCI_DEVICE_CLASS(class);
c->realize = e1000e_pci_realize;
@@ -679,8 +680,9 @@ static void e1000e_class_init(ObjectClass *class, void *data)
c->romfile = "efi-e1000e.rom";
c->class_id = PCI_CLASS_NETWORK_ETHERNET;
+ rc->phases.hold = e1000e_qdev_reset_hold;
+
dc->desc = "Intel 82574L GbE Controller";
- dc->reset = e1000e_qdev_reset;
dc->vmsd = &e1000e_vmstate;
e1000e_prop_disable_vnet = qdev_prop_uint8;
diff --git a/hw/net/trace-events b/hw/net/trace-events
index 8fa4299704..c98ad12537 100644
--- a/hw/net/trace-events
+++ b/hw/net/trace-events
@@ -251,7 +251,7 @@ e1000e_vm_state_stopped(void) "VM state is stopped"
# e1000e.c
e1000e_cb_pci_realize(void) "E1000E PCI realize entry"
e1000e_cb_pci_uninit(void) "E1000E PCI unit entry"
-e1000e_cb_qdev_reset(void) "E1000E qdev reset entry"
+e1000e_cb_qdev_reset_hold(void) "E1000E qdev reset hold"
e1000e_cb_pre_save(void) "E1000E pre save entry"
e1000e_cb_post_load(void) "E1000E post load entry"
--
2.39.1
^ permalink raw reply related [flat|nested] 33+ messages in thread* [PATCH v4 15/28] e1000e: Introduce e1000_rx_desc_union
2023-01-30 13:46 [PATCH v4 00/28] e1000x cleanups (preliminary for IGB) Akihiko Odaki
` (13 preceding siblings ...)
2023-01-30 13:47 ` [PATCH v4 14/28] e1000e: " Akihiko Odaki
@ 2023-01-30 13:47 ` Akihiko Odaki
2023-01-30 13:47 ` [PATCH v4 16/28] e1000e: Set MII_ANER_NWAY Akihiko Odaki
` (12 subsequent siblings)
27 siblings, 0 replies; 33+ messages in thread
From: Akihiko Odaki @ 2023-01-30 13:47 UTC (permalink / raw)
Cc: Jason Wang, Dmitry Fleytman, Thomas Huth, Laurent Vivier,
Paolo Bonzini, qemu-devel, qemu-ppc, Michael S . Tsirkin,
Yan Vugenfirer, Yuri Benditovich, Sriram Yagnaraman,
Alexander Bulekov, Akihiko Odaki
Before this change, e1000e_write_packet_to_guest() allocated the
receive descriptor buffer as an array of uint8_t. This does not ensure
the buffer is sufficiently aligned.
Introduce e1000_rx_desc_union type, a union type of all receive
descriptor types to correct this.
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
---
hw/net/e1000_regs.h | 1 -
hw/net/e1000e_core.c | 115 +++++++++++++++++++++----------------------
2 files changed, 57 insertions(+), 59 deletions(-)
diff --git a/hw/net/e1000_regs.h b/hw/net/e1000_regs.h
index 6a36573802..4545fe25a6 100644
--- a/hw/net/e1000_regs.h
+++ b/hw/net/e1000_regs.h
@@ -1061,7 +1061,6 @@ union e1000_rx_desc_packet_split {
#define E1000_RING_DESC_LEN_SHIFT (4)
#define E1000_MIN_RX_DESC_LEN E1000_RING_DESC_LEN
-#define E1000_MAX_RX_DESC_LEN (sizeof(union e1000_rx_desc_packet_split))
/* Receive Descriptor bit definitions */
#define E1000_RXD_STAT_DD 0x01 /* Descriptor Done */
diff --git a/hw/net/e1000e_core.c b/hw/net/e1000e_core.c
index b8670662c8..d8c17baf8f 100644
--- a/hw/net/e1000e_core.c
+++ b/hw/net/e1000e_core.c
@@ -55,6 +55,12 @@
#define E1000E_MAX_TX_FRAGS (64)
+union e1000_rx_desc_union {
+ struct e1000_rx_desc legacy;
+ union e1000_rx_desc_extended extended;
+ union e1000_rx_desc_packet_split packet_split;
+};
+
static inline void
e1000e_set_interrupt_cause(E1000ECore *core, uint32_t val);
@@ -1053,29 +1059,28 @@ e1000e_receive_filter(E1000ECore *core, const uint8_t *buf, int size)
}
static inline void
-e1000e_read_lgcy_rx_descr(E1000ECore *core, uint8_t *desc, hwaddr *buff_addr)
+e1000e_read_lgcy_rx_descr(E1000ECore *core, struct e1000_rx_desc *desc,
+ hwaddr *buff_addr)
{
- struct e1000_rx_desc *d = (struct e1000_rx_desc *) desc;
- *buff_addr = le64_to_cpu(d->buffer_addr);
+ *buff_addr = le64_to_cpu(desc->buffer_addr);
}
static inline void
-e1000e_read_ext_rx_descr(E1000ECore *core, uint8_t *desc, hwaddr *buff_addr)
+e1000e_read_ext_rx_descr(E1000ECore *core, union e1000_rx_desc_extended *desc,
+ hwaddr *buff_addr)
{
- union e1000_rx_desc_extended *d = (union e1000_rx_desc_extended *) desc;
- *buff_addr = le64_to_cpu(d->read.buffer_addr);
+ *buff_addr = le64_to_cpu(desc->read.buffer_addr);
}
static inline void
-e1000e_read_ps_rx_descr(E1000ECore *core, uint8_t *desc,
+e1000e_read_ps_rx_descr(E1000ECore *core,
+ union e1000_rx_desc_packet_split *desc,
hwaddr (*buff_addr)[MAX_PS_BUFFERS])
{
int i;
- union e1000_rx_desc_packet_split *d =
- (union e1000_rx_desc_packet_split *) desc;
for (i = 0; i < MAX_PS_BUFFERS; i++) {
- (*buff_addr)[i] = le64_to_cpu(d->read.buffer_addr[i]);
+ (*buff_addr)[i] = le64_to_cpu(desc->read.buffer_addr[i]);
}
trace_e1000e_rx_desc_ps_read((*buff_addr)[0], (*buff_addr)[1],
@@ -1083,17 +1088,17 @@ e1000e_read_ps_rx_descr(E1000ECore *core, uint8_t *desc,
}
static inline void
-e1000e_read_rx_descr(E1000ECore *core, uint8_t *desc,
+e1000e_read_rx_descr(E1000ECore *core, union e1000_rx_desc_union *desc,
hwaddr (*buff_addr)[MAX_PS_BUFFERS])
{
if (e1000e_rx_use_legacy_descriptor(core)) {
- e1000e_read_lgcy_rx_descr(core, desc, &(*buff_addr)[0]);
+ e1000e_read_lgcy_rx_descr(core, &desc->legacy, &(*buff_addr)[0]);
(*buff_addr)[1] = (*buff_addr)[2] = (*buff_addr)[3] = 0;
} else {
if (core->mac[RCTL] & E1000_RCTL_DTYP_PS) {
- e1000e_read_ps_rx_descr(core, desc, buff_addr);
+ e1000e_read_ps_rx_descr(core, &desc->packet_split, buff_addr);
} else {
- e1000e_read_ext_rx_descr(core, desc, &(*buff_addr)[0]);
+ e1000e_read_ext_rx_descr(core, &desc->extended, &(*buff_addr)[0]);
(*buff_addr)[1] = (*buff_addr)[2] = (*buff_addr)[3] = 0;
}
}
@@ -1264,7 +1269,7 @@ func_exit:
}
static inline void
-e1000e_write_lgcy_rx_descr(E1000ECore *core, uint8_t *desc,
+e1000e_write_lgcy_rx_descr(E1000ECore *core, struct e1000_rx_desc *desc,
struct NetRxPkt *pkt,
const E1000E_RSSInfo *rss_info,
uint16_t length)
@@ -1272,71 +1277,66 @@ e1000e_write_lgcy_rx_descr(E1000ECore *core, uint8_t *desc,
uint32_t status_flags, rss, mrq;
uint16_t ip_id;
- struct e1000_rx_desc *d = (struct e1000_rx_desc *) desc;
-
assert(!rss_info->enabled);
- d->length = cpu_to_le16(length);
- d->csum = 0;
+ desc->length = cpu_to_le16(length);
+ desc->csum = 0;
e1000e_build_rx_metadata(core, pkt, pkt != NULL,
rss_info,
&rss, &mrq,
&status_flags, &ip_id,
- &d->special);
- d->errors = (uint8_t) (le32_to_cpu(status_flags) >> 24);
- d->status = (uint8_t) le32_to_cpu(status_flags);
+ &desc->special);
+ desc->errors = (uint8_t) (le32_to_cpu(status_flags) >> 24);
+ desc->status = (uint8_t) le32_to_cpu(status_flags);
}
static inline void
-e1000e_write_ext_rx_descr(E1000ECore *core, uint8_t *desc,
+e1000e_write_ext_rx_descr(E1000ECore *core, union e1000_rx_desc_extended *desc,
struct NetRxPkt *pkt,
const E1000E_RSSInfo *rss_info,
uint16_t length)
{
- union e1000_rx_desc_extended *d = (union e1000_rx_desc_extended *) desc;
-
- memset(&d->wb, 0, sizeof(d->wb));
+ memset(&desc->wb, 0, sizeof(desc->wb));
- d->wb.upper.length = cpu_to_le16(length);
+ desc->wb.upper.length = cpu_to_le16(length);
e1000e_build_rx_metadata(core, pkt, pkt != NULL,
rss_info,
- &d->wb.lower.hi_dword.rss,
- &d->wb.lower.mrq,
- &d->wb.upper.status_error,
- &d->wb.lower.hi_dword.csum_ip.ip_id,
- &d->wb.upper.vlan);
+ &desc->wb.lower.hi_dword.rss,
+ &desc->wb.lower.mrq,
+ &desc->wb.upper.status_error,
+ &desc->wb.lower.hi_dword.csum_ip.ip_id,
+ &desc->wb.upper.vlan);
}
static inline void
-e1000e_write_ps_rx_descr(E1000ECore *core, uint8_t *desc,
+e1000e_write_ps_rx_descr(E1000ECore *core,
+ union e1000_rx_desc_packet_split *desc,
struct NetRxPkt *pkt,
const E1000E_RSSInfo *rss_info,
size_t ps_hdr_len,
uint16_t(*written)[MAX_PS_BUFFERS])
{
int i;
- union e1000_rx_desc_packet_split *d =
- (union e1000_rx_desc_packet_split *) desc;
- memset(&d->wb, 0, sizeof(d->wb));
+ memset(&desc->wb, 0, sizeof(desc->wb));
- d->wb.middle.length0 = cpu_to_le16((*written)[0]);
+ desc->wb.middle.length0 = cpu_to_le16((*written)[0]);
for (i = 0; i < PS_PAGE_BUFFERS; i++) {
- d->wb.upper.length[i] = cpu_to_le16((*written)[i + 1]);
+ desc->wb.upper.length[i] = cpu_to_le16((*written)[i + 1]);
}
e1000e_build_rx_metadata(core, pkt, pkt != NULL,
rss_info,
- &d->wb.lower.hi_dword.rss,
- &d->wb.lower.mrq,
- &d->wb.middle.status_error,
- &d->wb.lower.hi_dword.csum_ip.ip_id,
- &d->wb.middle.vlan);
+ &desc->wb.lower.hi_dword.rss,
+ &desc->wb.lower.mrq,
+ &desc->wb.middle.status_error,
+ &desc->wb.lower.hi_dword.csum_ip.ip_id,
+ &desc->wb.middle.vlan);
- d->wb.upper.header_status =
+ desc->wb.upper.header_status =
cpu_to_le16(ps_hdr_len | (ps_hdr_len ? E1000_RXDPS_HDRSTAT_HDRSP : 0));
trace_e1000e_rx_desc_ps_write((*written)[0], (*written)[1],
@@ -1344,20 +1344,21 @@ e1000e_write_ps_rx_descr(E1000ECore *core, uint8_t *desc,
}
static inline void
-e1000e_write_rx_descr(E1000ECore *core, uint8_t *desc,
+e1000e_write_rx_descr(E1000ECore *core, union e1000_rx_desc_union *desc,
struct NetRxPkt *pkt, const E1000E_RSSInfo *rss_info,
size_t ps_hdr_len, uint16_t(*written)[MAX_PS_BUFFERS])
{
if (e1000e_rx_use_legacy_descriptor(core)) {
assert(ps_hdr_len == 0);
- e1000e_write_lgcy_rx_descr(core, desc, pkt, rss_info, (*written)[0]);
+ e1000e_write_lgcy_rx_descr(core, &desc->legacy, pkt, rss_info,
+ (*written)[0]);
} else {
if (core->mac[RCTL] & E1000_RCTL_DTYP_PS) {
- e1000e_write_ps_rx_descr(core, desc, pkt, rss_info,
+ e1000e_write_ps_rx_descr(core, &desc->packet_split, pkt, rss_info,
ps_hdr_len, written);
} else {
assert(ps_hdr_len == 0);
- e1000e_write_ext_rx_descr(core, desc, pkt, rss_info,
+ e1000e_write_ext_rx_descr(core, &desc->extended, pkt, rss_info,
(*written)[0]);
}
}
@@ -1365,12 +1366,12 @@ struct NetRxPkt *pkt, const E1000E_RSSInfo *rss_info,
static inline void
e1000e_pci_dma_write_rx_desc(E1000ECore *core, dma_addr_t addr,
- uint8_t *desc, dma_addr_t len)
+ union e1000_rx_desc_union *desc, dma_addr_t len)
{
PCIDevice *dev = core->owner;
if (e1000e_rx_use_legacy_descriptor(core)) {
- struct e1000_rx_desc *d = (struct e1000_rx_desc *) desc;
+ struct e1000_rx_desc *d = &desc->legacy;
size_t offset = offsetof(struct e1000_rx_desc, status);
uint8_t status = d->status;
@@ -1383,8 +1384,7 @@ e1000e_pci_dma_write_rx_desc(E1000ECore *core, dma_addr_t addr,
}
} else {
if (core->mac[RCTL] & E1000_RCTL_DTYP_PS) {
- union e1000_rx_desc_packet_split *d =
- (union e1000_rx_desc_packet_split *) desc;
+ union e1000_rx_desc_packet_split *d = &desc->packet_split;
size_t offset = offsetof(union e1000_rx_desc_packet_split,
wb.middle.status_error);
uint32_t status = d->wb.middle.status_error;
@@ -1397,8 +1397,7 @@ e1000e_pci_dma_write_rx_desc(E1000ECore *core, dma_addr_t addr,
pci_dma_write(dev, addr + offset, &status, sizeof(status));
}
} else {
- union e1000_rx_desc_extended *d =
- (union e1000_rx_desc_extended *) desc;
+ union e1000_rx_desc_extended *d = &desc->extended;
size_t offset = offsetof(union e1000_rx_desc_extended,
wb.upper.status_error);
uint32_t status = d->wb.upper.status_error;
@@ -1542,7 +1541,7 @@ e1000e_write_packet_to_guest(E1000ECore *core, struct NetRxPkt *pkt,
{
PCIDevice *d = core->owner;
dma_addr_t base;
- uint8_t desc[E1000_MAX_RX_DESC_LEN];
+ union e1000_rx_desc_union desc;
size_t desc_size;
size_t desc_offset = 0;
size_t iov_ofs = 0;
@@ -1578,7 +1577,7 @@ e1000e_write_packet_to_guest(E1000ECore *core, struct NetRxPkt *pkt,
trace_e1000e_rx_descr(rxi->idx, base, core->rx_desc_len);
- e1000e_read_rx_descr(core, desc, &ba);
+ e1000e_read_rx_descr(core, &desc, &ba);
if (ba[0]) {
if (desc_offset < size) {
@@ -1648,9 +1647,9 @@ e1000e_write_packet_to_guest(E1000ECore *core, struct NetRxPkt *pkt,
is_last = true;
}
- e1000e_write_rx_descr(core, desc, is_last ? core->rx_pkt : NULL,
+ e1000e_write_rx_descr(core, &desc, is_last ? core->rx_pkt : NULL,
rss_info, do_ps ? ps_hdr_len : 0, &bastate.written);
- e1000e_pci_dma_write_rx_desc(core, base, desc, core->rx_desc_len);
+ e1000e_pci_dma_write_rx_desc(core, base, &desc, core->rx_desc_len);
e1000e_ring_advance(core, rxi,
core->rx_desc_len / E1000_MIN_RX_DESC_LEN);
--
2.39.1
^ permalink raw reply related [flat|nested] 33+ messages in thread* [PATCH v4 16/28] e1000e: Set MII_ANER_NWAY
2023-01-30 13:46 [PATCH v4 00/28] e1000x cleanups (preliminary for IGB) Akihiko Odaki
` (14 preceding siblings ...)
2023-01-30 13:47 ` [PATCH v4 15/28] e1000e: Introduce e1000_rx_desc_union Akihiko Odaki
@ 2023-01-30 13:47 ` Akihiko Odaki
2023-01-30 13:47 ` [PATCH v4 17/28] e1000e: Remove extra pointer indirection Akihiko Odaki
` (11 subsequent siblings)
27 siblings, 0 replies; 33+ messages in thread
From: Akihiko Odaki @ 2023-01-30 13:47 UTC (permalink / raw)
Cc: Jason Wang, Dmitry Fleytman, Thomas Huth, Laurent Vivier,
Paolo Bonzini, qemu-devel, qemu-ppc, Michael S . Tsirkin,
Yan Vugenfirer, Yuri Benditovich, Sriram Yagnaraman,
Alexander Bulekov, Akihiko Odaki, Philippe Mathieu-Daudé
This keeps Windows driver 12.18.9.23 from generating an event with ID
30. The description of the event is as follows:
> Intel(R) 82574L Gigabit Network Connection
> PROBLEM: The network adapter is configured for auto-negotiation but
> the link partner is not. This may result in a duplex mismatch.
> ACTION: Configure the link partner for auto-negotiation.
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
---
hw/net/e1000e_core.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/hw/net/e1000e_core.c b/hw/net/e1000e_core.c
index d8c17baf8f..736708407c 100644
--- a/hw/net/e1000e_core.c
+++ b/hw/net/e1000e_core.c
@@ -3426,7 +3426,7 @@ e1000e_phy_reg_init[E1000E_PHY_PAGES][E1000E_PHY_PAGE_SIZE] = {
[MII_ANLPAR] = MII_ANLPAR_10 | MII_ANLPAR_10FD |
MII_ANLPAR_TX | MII_ANLPAR_TXFD |
MII_ANLPAR_T4 | MII_ANLPAR_PAUSE,
- [MII_ANER] = MII_ANER_NP,
+ [MII_ANER] = MII_ANER_NP | MII_ANER_NWAY,
[MII_ANNP] = 1 | MII_ANNP_MP,
[MII_CTRL1000] = MII_CTRL1000_HALF | MII_CTRL1000_FULL |
MII_CTRL1000_PORT | MII_CTRL1000_MASTER,
--
2.39.1
^ permalink raw reply related [flat|nested] 33+ messages in thread* [PATCH v4 17/28] e1000e: Remove extra pointer indirection
2023-01-30 13:46 [PATCH v4 00/28] e1000x cleanups (preliminary for IGB) Akihiko Odaki
` (15 preceding siblings ...)
2023-01-30 13:47 ` [PATCH v4 16/28] e1000e: Set MII_ANER_NWAY Akihiko Odaki
@ 2023-01-30 13:47 ` Akihiko Odaki
2023-01-30 13:47 ` [PATCH v4 18/28] net: Check L4 header size Akihiko Odaki
` (10 subsequent siblings)
27 siblings, 0 replies; 33+ messages in thread
From: Akihiko Odaki @ 2023-01-30 13:47 UTC (permalink / raw)
Cc: Jason Wang, Dmitry Fleytman, Thomas Huth, Laurent Vivier,
Paolo Bonzini, qemu-devel, qemu-ppc, Michael S . Tsirkin,
Yan Vugenfirer, Yuri Benditovich, Sriram Yagnaraman,
Alexander Bulekov, Akihiko Odaki, Philippe Mathieu-Daudé
e1000e_write_packet_to_guest() passes the reference of variable ba as a
pointer to an array, and that pointer indirection is just unnecessary;
all functions which uses the passed reference performs no pointer
operation on the pointer and they simply dereference the passed
pointer. Remove the extra pointer indirection.
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
---
hw/net/e1000e_core.c | 38 +++++++++++++++++++-------------------
1 file changed, 19 insertions(+), 19 deletions(-)
diff --git a/hw/net/e1000e_core.c b/hw/net/e1000e_core.c
index 736708407c..d143f2ae6f 100644
--- a/hw/net/e1000e_core.c
+++ b/hw/net/e1000e_core.c
@@ -1075,31 +1075,31 @@ e1000e_read_ext_rx_descr(E1000ECore *core, union e1000_rx_desc_extended *desc,
static inline void
e1000e_read_ps_rx_descr(E1000ECore *core,
union e1000_rx_desc_packet_split *desc,
- hwaddr (*buff_addr)[MAX_PS_BUFFERS])
+ hwaddr buff_addr[MAX_PS_BUFFERS])
{
int i;
for (i = 0; i < MAX_PS_BUFFERS; i++) {
- (*buff_addr)[i] = le64_to_cpu(desc->read.buffer_addr[i]);
+ buff_addr[i] = le64_to_cpu(desc->read.buffer_addr[i]);
}
- trace_e1000e_rx_desc_ps_read((*buff_addr)[0], (*buff_addr)[1],
- (*buff_addr)[2], (*buff_addr)[3]);
+ trace_e1000e_rx_desc_ps_read(buff_addr[0], buff_addr[1],
+ buff_addr[2], buff_addr[3]);
}
static inline void
e1000e_read_rx_descr(E1000ECore *core, union e1000_rx_desc_union *desc,
- hwaddr (*buff_addr)[MAX_PS_BUFFERS])
+ hwaddr buff_addr[MAX_PS_BUFFERS])
{
if (e1000e_rx_use_legacy_descriptor(core)) {
- e1000e_read_lgcy_rx_descr(core, &desc->legacy, &(*buff_addr)[0]);
- (*buff_addr)[1] = (*buff_addr)[2] = (*buff_addr)[3] = 0;
+ e1000e_read_lgcy_rx_descr(core, &desc->legacy, &buff_addr[0]);
+ buff_addr[1] = buff_addr[2] = buff_addr[3] = 0;
} else {
if (core->mac[RCTL] & E1000_RCTL_DTYP_PS) {
e1000e_read_ps_rx_descr(core, &desc->packet_split, buff_addr);
} else {
- e1000e_read_ext_rx_descr(core, &desc->extended, &(*buff_addr)[0]);
- (*buff_addr)[1] = (*buff_addr)[2] = (*buff_addr)[3] = 0;
+ e1000e_read_ext_rx_descr(core, &desc->extended, &buff_addr[0]);
+ buff_addr[1] = buff_addr[2] = buff_addr[3] = 0;
}
}
}
@@ -1420,14 +1420,14 @@ typedef struct e1000e_ba_state_st {
static inline void
e1000e_write_hdr_to_rx_buffers(E1000ECore *core,
- hwaddr (*ba)[MAX_PS_BUFFERS],
+ hwaddr ba[MAX_PS_BUFFERS],
e1000e_ba_state *bastate,
const char *data,
dma_addr_t data_len)
{
assert(data_len <= core->rxbuf_sizes[0] - bastate->written[0]);
- pci_dma_write(core->owner, (*ba)[0] + bastate->written[0], data, data_len);
+ pci_dma_write(core->owner, ba[0] + bastate->written[0], data, data_len);
bastate->written[0] += data_len;
bastate->cur_idx = 1;
@@ -1435,7 +1435,7 @@ e1000e_write_hdr_to_rx_buffers(E1000ECore *core,
static void
e1000e_write_to_rx_buffers(E1000ECore *core,
- hwaddr (*ba)[MAX_PS_BUFFERS],
+ hwaddr ba[MAX_PS_BUFFERS],
e1000e_ba_state *bastate,
const char *data,
dma_addr_t data_len)
@@ -1447,13 +1447,13 @@ e1000e_write_to_rx_buffers(E1000ECore *core,
uint32_t bytes_to_write = MIN(data_len, cur_buf_bytes_left);
trace_e1000e_rx_desc_buff_write(bastate->cur_idx,
- (*ba)[bastate->cur_idx],
+ ba[bastate->cur_idx],
bastate->written[bastate->cur_idx],
data,
bytes_to_write);
pci_dma_write(core->owner,
- (*ba)[bastate->cur_idx] + bastate->written[bastate->cur_idx],
+ ba[bastate->cur_idx] + bastate->written[bastate->cur_idx],
data, bytes_to_write);
bastate->written[bastate->cur_idx] += bytes_to_write;
@@ -1577,7 +1577,7 @@ e1000e_write_packet_to_guest(E1000ECore *core, struct NetRxPkt *pkt,
trace_e1000e_rx_descr(rxi->idx, base, core->rx_desc_len);
- e1000e_read_rx_descr(core, &desc, &ba);
+ e1000e_read_rx_descr(core, &desc, ba);
if (ba[0]) {
if (desc_offset < size) {
@@ -1596,7 +1596,7 @@ e1000e_write_packet_to_guest(E1000ECore *core, struct NetRxPkt *pkt,
iov_copy = MIN(ps_hdr_len - ps_hdr_copied,
iov->iov_len - iov_ofs);
- e1000e_write_hdr_to_rx_buffers(core, &ba, &bastate,
+ e1000e_write_hdr_to_rx_buffers(core, ba, &bastate,
iov->iov_base, iov_copy);
copy_size -= iov_copy;
@@ -1613,7 +1613,7 @@ e1000e_write_packet_to_guest(E1000ECore *core, struct NetRxPkt *pkt,
} else {
/* Leave buffer 0 of each descriptor except first */
/* empty as per spec 7.1.5.1 */
- e1000e_write_hdr_to_rx_buffers(core, &ba, &bastate,
+ e1000e_write_hdr_to_rx_buffers(core, ba, &bastate,
NULL, 0);
}
}
@@ -1622,7 +1622,7 @@ e1000e_write_packet_to_guest(E1000ECore *core, struct NetRxPkt *pkt,
while (copy_size) {
iov_copy = MIN(copy_size, iov->iov_len - iov_ofs);
- e1000e_write_to_rx_buffers(core, &ba, &bastate,
+ e1000e_write_to_rx_buffers(core, ba, &bastate,
iov->iov_base + iov_ofs, iov_copy);
copy_size -= iov_copy;
@@ -1635,7 +1635,7 @@ e1000e_write_packet_to_guest(E1000ECore *core, struct NetRxPkt *pkt,
if (desc_offset + desc_size >= total_size) {
/* Simulate FCS checksum presence in the last descriptor */
- e1000e_write_to_rx_buffers(core, &ba, &bastate,
+ e1000e_write_to_rx_buffers(core, ba, &bastate,
(const char *) &fcs_pad, e1000x_fcs_len(core->mac));
}
}
--
2.39.1
^ permalink raw reply related [flat|nested] 33+ messages in thread* [PATCH v4 18/28] net: Check L4 header size
2023-01-30 13:46 [PATCH v4 00/28] e1000x cleanups (preliminary for IGB) Akihiko Odaki
` (16 preceding siblings ...)
2023-01-30 13:47 ` [PATCH v4 17/28] e1000e: Remove extra pointer indirection Akihiko Odaki
@ 2023-01-30 13:47 ` Akihiko Odaki
2023-01-30 13:47 ` [PATCH v4 19/28] e1000x: Alter the signature of e1000x_is_vlan_packet Akihiko Odaki
` (9 subsequent siblings)
27 siblings, 0 replies; 33+ messages in thread
From: Akihiko Odaki @ 2023-01-30 13:47 UTC (permalink / raw)
Cc: Jason Wang, Dmitry Fleytman, Thomas Huth, Laurent Vivier,
Paolo Bonzini, qemu-devel, qemu-ppc, Michael S . Tsirkin,
Yan Vugenfirer, Yuri Benditovich, Sriram Yagnaraman,
Alexander Bulekov, Akihiko Odaki
net_tx_pkt_build_vheader() inspects TCP header but had no check for
the header size, resulting in an undefined behavior. Check the header
size and drop the packet if the header is too small.
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
---
hw/net/e1000e_core.c | 19 ++++++++++++++-----
hw/net/net_tx_pkt.c | 13 ++++++++++---
hw/net/net_tx_pkt.h | 3 ++-
hw/net/vmxnet3.c | 14 +++++++-------
4 files changed, 33 insertions(+), 16 deletions(-)
diff --git a/hw/net/e1000e_core.c b/hw/net/e1000e_core.c
index d143f2ae6f..38d374fba3 100644
--- a/hw/net/e1000e_core.c
+++ b/hw/net/e1000e_core.c
@@ -629,23 +629,30 @@ e1000e_rss_parse_packet(E1000ECore *core,
info->queue = E1000_RSS_QUEUE(&core->mac[RETA], info->hash);
}
-static void
+static bool
e1000e_setup_tx_offloads(E1000ECore *core, struct e1000e_tx *tx)
{
if (tx->props.tse && tx->cptse) {
- net_tx_pkt_build_vheader(tx->tx_pkt, true, true, tx->props.mss);
+ if (!net_tx_pkt_build_vheader(tx->tx_pkt, true, true, tx->props.mss)) {
+ return false;
+ }
+
net_tx_pkt_update_ip_checksums(tx->tx_pkt);
e1000x_inc_reg_if_not_full(core->mac, TSCTC);
- return;
+ return true;
}
if (tx->sum_needed & E1000_TXD_POPTS_TXSM) {
- net_tx_pkt_build_vheader(tx->tx_pkt, false, true, 0);
+ if (!net_tx_pkt_build_vheader(tx->tx_pkt, false, true, 0)) {
+ return false;
+ }
}
if (tx->sum_needed & E1000_TXD_POPTS_IXSM) {
net_tx_pkt_update_ip_hdr_checksum(tx->tx_pkt);
}
+
+ return true;
}
static bool
@@ -654,7 +661,9 @@ e1000e_tx_pkt_send(E1000ECore *core, struct e1000e_tx *tx, int queue_index)
int target_queue = MIN(core->max_queue_num, queue_index);
NetClientState *queue = qemu_get_subqueue(core->owner_nic, target_queue);
- e1000e_setup_tx_offloads(core, tx);
+ if (!e1000e_setup_tx_offloads(core, tx)) {
+ return false;
+ }
net_tx_pkt_dump(tx->tx_pkt);
diff --git a/hw/net/net_tx_pkt.c b/hw/net/net_tx_pkt.c
index 2533ea2700..8a23899a4d 100644
--- a/hw/net/net_tx_pkt.c
+++ b/hw/net/net_tx_pkt.c
@@ -304,10 +304,11 @@ func_exit:
return rc;
}
-void net_tx_pkt_build_vheader(struct NetTxPkt *pkt, bool tso_enable,
+bool net_tx_pkt_build_vheader(struct NetTxPkt *pkt, bool tso_enable,
bool csum_enable, uint32_t gso_size)
{
struct tcp_hdr l4hdr;
+ size_t bytes_read;
assert(pkt);
/* csum has to be enabled if tso is. */
@@ -328,8 +329,12 @@ void net_tx_pkt_build_vheader(struct NetTxPkt *pkt, bool tso_enable,
case VIRTIO_NET_HDR_GSO_TCPV4:
case VIRTIO_NET_HDR_GSO_TCPV6:
- iov_to_buf(&pkt->vec[NET_TX_PKT_PL_START_FRAG], pkt->payload_frags,
- 0, &l4hdr, sizeof(l4hdr));
+ bytes_read = iov_to_buf(&pkt->vec[NET_TX_PKT_PL_START_FRAG],
+ pkt->payload_frags, 0, &l4hdr, sizeof(l4hdr));
+ if (bytes_read < sizeof(l4hdr)) {
+ return false;
+ }
+
pkt->virt_hdr.hdr_len = pkt->hdr_len + l4hdr.th_off * sizeof(uint32_t);
pkt->virt_hdr.gso_size = gso_size;
break;
@@ -354,6 +359,8 @@ void net_tx_pkt_build_vheader(struct NetTxPkt *pkt, bool tso_enable,
break;
}
}
+
+ return true;
}
void net_tx_pkt_setup_vlan_header_ex(struct NetTxPkt *pkt,
diff --git a/hw/net/net_tx_pkt.h b/hw/net/net_tx_pkt.h
index 4ec8bbe9bd..2e38a5fa69 100644
--- a/hw/net/net_tx_pkt.h
+++ b/hw/net/net_tx_pkt.h
@@ -59,9 +59,10 @@ struct virtio_net_hdr *net_tx_pkt_get_vhdr(struct NetTxPkt *pkt);
* @tso_enable: TSO enabled
* @csum_enable: CSO enabled
* @gso_size: MSS size for TSO
+ * @ret: operation result
*
*/
-void net_tx_pkt_build_vheader(struct NetTxPkt *pkt, bool tso_enable,
+bool net_tx_pkt_build_vheader(struct NetTxPkt *pkt, bool tso_enable,
bool csum_enable, uint32_t gso_size);
/**
diff --git a/hw/net/vmxnet3.c b/hw/net/vmxnet3.c
index d2ab527ef4..c63bbb59bd 100644
--- a/hw/net/vmxnet3.c
+++ b/hw/net/vmxnet3.c
@@ -440,19 +440,19 @@ vmxnet3_setup_tx_offloads(VMXNET3State *s)
{
switch (s->offload_mode) {
case VMXNET3_OM_NONE:
- net_tx_pkt_build_vheader(s->tx_pkt, false, false, 0);
- break;
+ return net_tx_pkt_build_vheader(s->tx_pkt, false, false, 0);
case VMXNET3_OM_CSUM:
- net_tx_pkt_build_vheader(s->tx_pkt, false, true, 0);
VMW_PKPRN("L4 CSO requested\n");
- break;
+ return net_tx_pkt_build_vheader(s->tx_pkt, false, true, 0);
case VMXNET3_OM_TSO:
- net_tx_pkt_build_vheader(s->tx_pkt, true, true,
- s->cso_or_gso_size);
- net_tx_pkt_update_ip_checksums(s->tx_pkt);
VMW_PKPRN("GSO offload requested.");
+ if (!net_tx_pkt_build_vheader(s->tx_pkt, true, true,
+ s->cso_or_gso_size)) {
+ return false;
+ }
+ net_tx_pkt_update_ip_checksums(s->tx_pkt);
break;
default:
--
2.39.1
^ permalink raw reply related [flat|nested] 33+ messages in thread* [PATCH v4 19/28] e1000x: Alter the signature of e1000x_is_vlan_packet
2023-01-30 13:46 [PATCH v4 00/28] e1000x cleanups (preliminary for IGB) Akihiko Odaki
` (17 preceding siblings ...)
2023-01-30 13:47 ` [PATCH v4 18/28] net: Check L4 header size Akihiko Odaki
@ 2023-01-30 13:47 ` Akihiko Odaki
2023-01-30 13:47 ` [PATCH v4 20/28] net: Strip virtio-net header when dumping Akihiko Odaki
` (8 subsequent siblings)
27 siblings, 0 replies; 33+ messages in thread
From: Akihiko Odaki @ 2023-01-30 13:47 UTC (permalink / raw)
Cc: Jason Wang, Dmitry Fleytman, Thomas Huth, Laurent Vivier,
Paolo Bonzini, qemu-devel, qemu-ppc, Michael S . Tsirkin,
Yan Vugenfirer, Yuri Benditovich, Sriram Yagnaraman,
Alexander Bulekov, Akihiko Odaki
e1000x_is_vlan_packet() had a pointer to uint8_t as a parameter, but
it does not have to be uint8_t. Change the type to void *.
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
---
hw/net/e1000x_common.c | 2 +-
hw/net/e1000x_common.h | 2 +-
2 files changed, 2 insertions(+), 2 deletions(-)
diff --git a/hw/net/e1000x_common.c b/hw/net/e1000x_common.c
index b3bbf31582..e6387dde53 100644
--- a/hw/net/e1000x_common.c
+++ b/hw/net/e1000x_common.c
@@ -47,7 +47,7 @@ bool e1000x_rx_ready(PCIDevice *d, uint32_t *mac)
return true;
}
-bool e1000x_is_vlan_packet(const uint8_t *buf, uint16_t vet)
+bool e1000x_is_vlan_packet(const void *buf, uint16_t vet)
{
uint16_t eth_proto = lduw_be_p(&PKT_GET_ETH_HDR(buf)->h_proto);
bool res = (eth_proto == vet);
diff --git a/hw/net/e1000x_common.h b/hw/net/e1000x_common.h
index b991d814b1..86a31b69f8 100644
--- a/hw/net/e1000x_common.h
+++ b/hw/net/e1000x_common.h
@@ -178,7 +178,7 @@ uint32_t e1000x_rxbufsize(uint32_t rctl);
bool e1000x_rx_ready(PCIDevice *d, uint32_t *mac);
-bool e1000x_is_vlan_packet(const uint8_t *buf, uint16_t vet);
+bool e1000x_is_vlan_packet(const void *buf, uint16_t vet);
bool e1000x_rx_group_filter(uint32_t *mac, const uint8_t *buf);
--
2.39.1
^ permalink raw reply related [flat|nested] 33+ messages in thread* [PATCH v4 20/28] net: Strip virtio-net header when dumping
2023-01-30 13:46 [PATCH v4 00/28] e1000x cleanups (preliminary for IGB) Akihiko Odaki
` (18 preceding siblings ...)
2023-01-30 13:47 ` [PATCH v4 19/28] e1000x: Alter the signature of e1000x_is_vlan_packet Akihiko Odaki
@ 2023-01-30 13:47 ` Akihiko Odaki
2023-01-30 15:12 ` Michael S. Tsirkin
2023-01-30 13:47 ` [PATCH v4 21/28] hw/net/net_tx_pkt: Automatically determine if virtio-net header is used Akihiko Odaki
` (7 subsequent siblings)
27 siblings, 1 reply; 33+ messages in thread
From: Akihiko Odaki @ 2023-01-30 13:47 UTC (permalink / raw)
Cc: Jason Wang, Dmitry Fleytman, Thomas Huth, Laurent Vivier,
Paolo Bonzini, qemu-devel, qemu-ppc, Michael S . Tsirkin,
Yan Vugenfirer, Yuri Benditovich, Sriram Yagnaraman,
Alexander Bulekov, Akihiko Odaki
filter-dump specifiees Ethernet as PCAP LinkType, which does not expect
virtio-net header. Having virtio-net header in such PCAP file breaks
PCAP unconsumable. Unfortunately currently there is no LinkType for
virtio-net so for now strip virtio-net header to convert the output to
Ethernet.
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
---
include/net/net.h | 6 ++++++
net/dump.c | 11 +++++++----
net/net.c | 18 ++++++++++++++++++
net/tap.c | 16 ++++++++++++++++
4 files changed, 47 insertions(+), 4 deletions(-)
diff --git a/include/net/net.h b/include/net/net.h
index dc20b31e9f..4b2d72b3fc 100644
--- a/include/net/net.h
+++ b/include/net/net.h
@@ -56,8 +56,10 @@ typedef RxFilterInfo *(QueryRxFilter)(NetClientState *);
typedef bool (HasUfo)(NetClientState *);
typedef bool (HasVnetHdr)(NetClientState *);
typedef bool (HasVnetHdrLen)(NetClientState *, int);
+typedef bool (GetUsingVnetHdr)(NetClientState *);
typedef void (UsingVnetHdr)(NetClientState *, bool);
typedef void (SetOffload)(NetClientState *, int, int, int, int, int);
+typedef int (GetVnetHdrLen)(NetClientState *);
typedef void (SetVnetHdrLen)(NetClientState *, int);
typedef int (SetVnetLE)(NetClientState *, bool);
typedef int (SetVnetBE)(NetClientState *, bool);
@@ -84,8 +86,10 @@ typedef struct NetClientInfo {
HasUfo *has_ufo;
HasVnetHdr *has_vnet_hdr;
HasVnetHdrLen *has_vnet_hdr_len;
+ GetUsingVnetHdr *get_using_vnet_hdr;
UsingVnetHdr *using_vnet_hdr;
SetOffload *set_offload;
+ GetVnetHdrLen *get_vnet_hdr_len;
SetVnetHdrLen *set_vnet_hdr_len;
SetVnetLE *set_vnet_le;
SetVnetBE *set_vnet_be;
@@ -183,9 +187,11 @@ void qemu_format_nic_info_str(NetClientState *nc, uint8_t macaddr[6]);
bool qemu_has_ufo(NetClientState *nc);
bool qemu_has_vnet_hdr(NetClientState *nc);
bool qemu_has_vnet_hdr_len(NetClientState *nc, int len);
+bool qemu_get_using_vnet_hdr(NetClientState *nc);
void qemu_using_vnet_hdr(NetClientState *nc, bool enable);
void qemu_set_offload(NetClientState *nc, int csum, int tso4, int tso6,
int ecn, int ufo);
+int qemu_get_vnet_hdr_len(NetClientState *nc);
void qemu_set_vnet_hdr_len(NetClientState *nc, int len);
int qemu_set_vnet_le(NetClientState *nc, bool is_le);
int qemu_set_vnet_be(NetClientState *nc, bool is_be);
diff --git a/net/dump.c b/net/dump.c
index 6a63b15359..7d05f16ca7 100644
--- a/net/dump.c
+++ b/net/dump.c
@@ -61,12 +61,13 @@ struct pcap_sf_pkthdr {
uint32_t len;
};
-static ssize_t dump_receive_iov(DumpState *s, const struct iovec *iov, int cnt)
+static ssize_t dump_receive_iov(DumpState *s, const struct iovec *iov, int cnt,
+ int offset)
{
struct pcap_sf_pkthdr hdr;
int64_t ts;
int caplen;
- size_t size = iov_size(iov, cnt);
+ size_t size = iov_size(iov, cnt) - offset;
struct iovec dumpiov[cnt + 1];
/* Early return in case of previous error. */
@@ -84,7 +85,7 @@ static ssize_t dump_receive_iov(DumpState *s, const struct iovec *iov, int cnt)
dumpiov[0].iov_base = &hdr;
dumpiov[0].iov_len = sizeof(hdr);
- cnt = iov_copy(&dumpiov[1], cnt, iov, cnt, 0, caplen);
+ cnt = iov_copy(&dumpiov[1], cnt, iov, cnt, offset, caplen);
if (writev(s->fd, dumpiov, cnt + 1) != sizeof(hdr) + caplen) {
error_report("network dump write error - stopping dump");
@@ -153,8 +154,10 @@ static ssize_t filter_dump_receive_iov(NetFilterState *nf, NetClientState *sndr,
int iovcnt, NetPacketSent *sent_cb)
{
NetFilterDumpState *nfds = FILTER_DUMP(nf);
+ int offset = qemu_get_using_vnet_hdr(nf->netdev) ?
+ qemu_get_vnet_hdr_len(nf->netdev) : 0;
- dump_receive_iov(&nfds->ds, iov, iovcnt);
+ dump_receive_iov(&nfds->ds, iov, iovcnt, offset);
return 0;
}
diff --git a/net/net.c b/net/net.c
index 2d01472998..03f17de5fc 100644
--- a/net/net.c
+++ b/net/net.c
@@ -513,6 +513,15 @@ bool qemu_has_vnet_hdr_len(NetClientState *nc, int len)
return nc->info->has_vnet_hdr_len(nc, len);
}
+bool qemu_get_using_vnet_hdr(NetClientState *nc)
+{
+ if (!nc || !nc->info->get_using_vnet_hdr) {
+ return false;
+ }
+
+ return nc->info->get_using_vnet_hdr(nc);
+}
+
void qemu_using_vnet_hdr(NetClientState *nc, bool enable)
{
if (!nc || !nc->info->using_vnet_hdr) {
@@ -532,6 +541,15 @@ void qemu_set_offload(NetClientState *nc, int csum, int tso4, int tso6,
nc->info->set_offload(nc, csum, tso4, tso6, ecn, ufo);
}
+int qemu_get_vnet_hdr_len(NetClientState *nc)
+{
+ if (!nc || !nc->info->get_vnet_hdr_len) {
+ return 0;
+ }
+
+ return nc->info->get_vnet_hdr_len(nc);
+}
+
void qemu_set_vnet_hdr_len(NetClientState *nc, int len)
{
if (!nc || !nc->info->set_vnet_hdr_len) {
diff --git a/net/tap.c b/net/tap.c
index 7d7bc1dc5f..1bf085d422 100644
--- a/net/tap.c
+++ b/net/tap.c
@@ -255,6 +255,13 @@ static bool tap_has_vnet_hdr_len(NetClientState *nc, int len)
return !!tap_probe_vnet_hdr_len(s->fd, len);
}
+static int tap_get_vnet_hdr_len(NetClientState *nc)
+{
+ TAPState *s = DO_UPCAST(TAPState, nc, nc);
+
+ return s->host_vnet_hdr_len;
+}
+
static void tap_set_vnet_hdr_len(NetClientState *nc, int len)
{
TAPState *s = DO_UPCAST(TAPState, nc, nc);
@@ -268,6 +275,13 @@ static void tap_set_vnet_hdr_len(NetClientState *nc, int len)
s->host_vnet_hdr_len = len;
}
+static bool tap_get_using_vnet_hdr(NetClientState *nc)
+{
+ TAPState *s = DO_UPCAST(TAPState, nc, nc);
+
+ return s->using_vnet_hdr;
+}
+
static void tap_using_vnet_hdr(NetClientState *nc, bool using_vnet_hdr)
{
TAPState *s = DO_UPCAST(TAPState, nc, nc);
@@ -372,8 +386,10 @@ static NetClientInfo net_tap_info = {
.has_ufo = tap_has_ufo,
.has_vnet_hdr = tap_has_vnet_hdr,
.has_vnet_hdr_len = tap_has_vnet_hdr_len,
+ .get_using_vnet_hdr = tap_get_using_vnet_hdr,
.using_vnet_hdr = tap_using_vnet_hdr,
.set_offload = tap_set_offload,
+ .get_vnet_hdr_len = tap_get_vnet_hdr_len,
.set_vnet_hdr_len = tap_set_vnet_hdr_len,
.set_vnet_le = tap_set_vnet_le,
.set_vnet_be = tap_set_vnet_be,
--
2.39.1
^ permalink raw reply related [flat|nested] 33+ messages in thread* Re: [PATCH v4 20/28] net: Strip virtio-net header when dumping
2023-01-30 13:47 ` [PATCH v4 20/28] net: Strip virtio-net header when dumping Akihiko Odaki
@ 2023-01-30 15:12 ` Michael S. Tsirkin
2023-01-30 15:36 ` Akihiko Odaki
0 siblings, 1 reply; 33+ messages in thread
From: Michael S. Tsirkin @ 2023-01-30 15:12 UTC (permalink / raw)
To: Akihiko Odaki
Cc: Jason Wang, Dmitry Fleytman, Thomas Huth, Laurent Vivier,
Paolo Bonzini, qemu-devel, qemu-ppc, Yan Vugenfirer,
Yuri Benditovich, Sriram Yagnaraman, Alexander Bulekov
On Mon, Jan 30, 2023 at 10:47:07PM +0900, Akihiko Odaki wrote:
> filter-dump specifiees Ethernet as PCAP LinkType, which does not expect
> virtio-net header. Having virtio-net header in such PCAP file breaks
> PCAP unconsumable. Unfortunately currently there is no LinkType for
> virtio-net so for now strip virtio-net header to convert the output to
> Ethernet.
>
> Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Probably means you need to calculate checksums and split gso too right?
> ---
> include/net/net.h | 6 ++++++
> net/dump.c | 11 +++++++----
> net/net.c | 18 ++++++++++++++++++
> net/tap.c | 16 ++++++++++++++++
> 4 files changed, 47 insertions(+), 4 deletions(-)
>
> diff --git a/include/net/net.h b/include/net/net.h
> index dc20b31e9f..4b2d72b3fc 100644
> --- a/include/net/net.h
> +++ b/include/net/net.h
> @@ -56,8 +56,10 @@ typedef RxFilterInfo *(QueryRxFilter)(NetClientState *);
> typedef bool (HasUfo)(NetClientState *);
> typedef bool (HasVnetHdr)(NetClientState *);
> typedef bool (HasVnetHdrLen)(NetClientState *, int);
> +typedef bool (GetUsingVnetHdr)(NetClientState *);
> typedef void (UsingVnetHdr)(NetClientState *, bool);
> typedef void (SetOffload)(NetClientState *, int, int, int, int, int);
> +typedef int (GetVnetHdrLen)(NetClientState *);
> typedef void (SetVnetHdrLen)(NetClientState *, int);
> typedef int (SetVnetLE)(NetClientState *, bool);
> typedef int (SetVnetBE)(NetClientState *, bool);
> @@ -84,8 +86,10 @@ typedef struct NetClientInfo {
> HasUfo *has_ufo;
> HasVnetHdr *has_vnet_hdr;
> HasVnetHdrLen *has_vnet_hdr_len;
> + GetUsingVnetHdr *get_using_vnet_hdr;
> UsingVnetHdr *using_vnet_hdr;
> SetOffload *set_offload;
> + GetVnetHdrLen *get_vnet_hdr_len;
> SetVnetHdrLen *set_vnet_hdr_len;
> SetVnetLE *set_vnet_le;
> SetVnetBE *set_vnet_be;
> @@ -183,9 +187,11 @@ void qemu_format_nic_info_str(NetClientState *nc, uint8_t macaddr[6]);
> bool qemu_has_ufo(NetClientState *nc);
> bool qemu_has_vnet_hdr(NetClientState *nc);
> bool qemu_has_vnet_hdr_len(NetClientState *nc, int len);
> +bool qemu_get_using_vnet_hdr(NetClientState *nc);
> void qemu_using_vnet_hdr(NetClientState *nc, bool enable);
> void qemu_set_offload(NetClientState *nc, int csum, int tso4, int tso6,
> int ecn, int ufo);
> +int qemu_get_vnet_hdr_len(NetClientState *nc);
> void qemu_set_vnet_hdr_len(NetClientState *nc, int len);
> int qemu_set_vnet_le(NetClientState *nc, bool is_le);
> int qemu_set_vnet_be(NetClientState *nc, bool is_be);
> diff --git a/net/dump.c b/net/dump.c
> index 6a63b15359..7d05f16ca7 100644
> --- a/net/dump.c
> +++ b/net/dump.c
> @@ -61,12 +61,13 @@ struct pcap_sf_pkthdr {
> uint32_t len;
> };
>
> -static ssize_t dump_receive_iov(DumpState *s, const struct iovec *iov, int cnt)
> +static ssize_t dump_receive_iov(DumpState *s, const struct iovec *iov, int cnt,
> + int offset)
> {
> struct pcap_sf_pkthdr hdr;
> int64_t ts;
> int caplen;
> - size_t size = iov_size(iov, cnt);
> + size_t size = iov_size(iov, cnt) - offset;
> struct iovec dumpiov[cnt + 1];
>
> /* Early return in case of previous error. */
> @@ -84,7 +85,7 @@ static ssize_t dump_receive_iov(DumpState *s, const struct iovec *iov, int cnt)
>
> dumpiov[0].iov_base = &hdr;
> dumpiov[0].iov_len = sizeof(hdr);
> - cnt = iov_copy(&dumpiov[1], cnt, iov, cnt, 0, caplen);
> + cnt = iov_copy(&dumpiov[1], cnt, iov, cnt, offset, caplen);
>
> if (writev(s->fd, dumpiov, cnt + 1) != sizeof(hdr) + caplen) {
> error_report("network dump write error - stopping dump");
> @@ -153,8 +154,10 @@ static ssize_t filter_dump_receive_iov(NetFilterState *nf, NetClientState *sndr,
> int iovcnt, NetPacketSent *sent_cb)
> {
> NetFilterDumpState *nfds = FILTER_DUMP(nf);
> + int offset = qemu_get_using_vnet_hdr(nf->netdev) ?
> + qemu_get_vnet_hdr_len(nf->netdev) : 0;
>
> - dump_receive_iov(&nfds->ds, iov, iovcnt);
> + dump_receive_iov(&nfds->ds, iov, iovcnt, offset);
> return 0;
> }
>
> diff --git a/net/net.c b/net/net.c
> index 2d01472998..03f17de5fc 100644
> --- a/net/net.c
> +++ b/net/net.c
> @@ -513,6 +513,15 @@ bool qemu_has_vnet_hdr_len(NetClientState *nc, int len)
> return nc->info->has_vnet_hdr_len(nc, len);
> }
>
> +bool qemu_get_using_vnet_hdr(NetClientState *nc)
> +{
> + if (!nc || !nc->info->get_using_vnet_hdr) {
> + return false;
> + }
> +
> + return nc->info->get_using_vnet_hdr(nc);
> +}
> +
> void qemu_using_vnet_hdr(NetClientState *nc, bool enable)
> {
> if (!nc || !nc->info->using_vnet_hdr) {
> @@ -532,6 +541,15 @@ void qemu_set_offload(NetClientState *nc, int csum, int tso4, int tso6,
> nc->info->set_offload(nc, csum, tso4, tso6, ecn, ufo);
> }
>
> +int qemu_get_vnet_hdr_len(NetClientState *nc)
> +{
> + if (!nc || !nc->info->get_vnet_hdr_len) {
> + return 0;
> + }
> +
> + return nc->info->get_vnet_hdr_len(nc);
> +}
> +
> void qemu_set_vnet_hdr_len(NetClientState *nc, int len)
> {
> if (!nc || !nc->info->set_vnet_hdr_len) {
> diff --git a/net/tap.c b/net/tap.c
> index 7d7bc1dc5f..1bf085d422 100644
> --- a/net/tap.c
> +++ b/net/tap.c
> @@ -255,6 +255,13 @@ static bool tap_has_vnet_hdr_len(NetClientState *nc, int len)
> return !!tap_probe_vnet_hdr_len(s->fd, len);
> }
>
> +static int tap_get_vnet_hdr_len(NetClientState *nc)
> +{
> + TAPState *s = DO_UPCAST(TAPState, nc, nc);
> +
> + return s->host_vnet_hdr_len;
> +}
> +
> static void tap_set_vnet_hdr_len(NetClientState *nc, int len)
> {
> TAPState *s = DO_UPCAST(TAPState, nc, nc);
> @@ -268,6 +275,13 @@ static void tap_set_vnet_hdr_len(NetClientState *nc, int len)
> s->host_vnet_hdr_len = len;
> }
>
> +static bool tap_get_using_vnet_hdr(NetClientState *nc)
> +{
> + TAPState *s = DO_UPCAST(TAPState, nc, nc);
> +
> + return s->using_vnet_hdr;
> +}
> +
> static void tap_using_vnet_hdr(NetClientState *nc, bool using_vnet_hdr)
> {
> TAPState *s = DO_UPCAST(TAPState, nc, nc);
> @@ -372,8 +386,10 @@ static NetClientInfo net_tap_info = {
> .has_ufo = tap_has_ufo,
> .has_vnet_hdr = tap_has_vnet_hdr,
> .has_vnet_hdr_len = tap_has_vnet_hdr_len,
> + .get_using_vnet_hdr = tap_get_using_vnet_hdr,
> .using_vnet_hdr = tap_using_vnet_hdr,
> .set_offload = tap_set_offload,
> + .get_vnet_hdr_len = tap_get_vnet_hdr_len,
> .set_vnet_hdr_len = tap_set_vnet_hdr_len,
> .set_vnet_le = tap_set_vnet_le,
> .set_vnet_be = tap_set_vnet_be,
> --
> 2.39.1
^ permalink raw reply [flat|nested] 33+ messages in thread* Re: [PATCH v4 20/28] net: Strip virtio-net header when dumping
2023-01-30 15:12 ` Michael S. Tsirkin
@ 2023-01-30 15:36 ` Akihiko Odaki
2023-01-30 15:47 ` Michael S. Tsirkin
0 siblings, 1 reply; 33+ messages in thread
From: Akihiko Odaki @ 2023-01-30 15:36 UTC (permalink / raw)
To: Michael S. Tsirkin
Cc: Jason Wang, Dmitry Fleytman, Thomas Huth, Laurent Vivier,
Paolo Bonzini, qemu-devel, qemu-ppc, Yan Vugenfirer,
Yuri Benditovich, Sriram Yagnaraman, Alexander Bulekov
On 2023/01/31 0:12, Michael S. Tsirkin wrote:
> On Mon, Jan 30, 2023 at 10:47:07PM +0900, Akihiko Odaki wrote:
>> filter-dump specifiees Ethernet as PCAP LinkType, which does not expect
>> virtio-net header. Having virtio-net header in such PCAP file breaks
>> PCAP unconsumable. Unfortunately currently there is no LinkType for
>> virtio-net so for now strip virtio-net header to convert the output to
>> Ethernet.
>>
>> Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
>
> Probably means you need to calculate checksums and split gso too right?
It was not necessary in my case as I used Wireshark and it tolerates
wrong checksums and large packets (it even says "Checksum incorrect
[maybe caused by 'TCP checksum offload'?]"). It was even more helpful to
have raw packets instead of transformed packets for debugging purposes.
Perhaps an option to transform packets may be added later if a need arises.
>
>> ---
>> include/net/net.h | 6 ++++++
>> net/dump.c | 11 +++++++----
>> net/net.c | 18 ++++++++++++++++++
>> net/tap.c | 16 ++++++++++++++++
>> 4 files changed, 47 insertions(+), 4 deletions(-)
>>
>> diff --git a/include/net/net.h b/include/net/net.h
>> index dc20b31e9f..4b2d72b3fc 100644
>> --- a/include/net/net.h
>> +++ b/include/net/net.h
>> @@ -56,8 +56,10 @@ typedef RxFilterInfo *(QueryRxFilter)(NetClientState *);
>> typedef bool (HasUfo)(NetClientState *);
>> typedef bool (HasVnetHdr)(NetClientState *);
>> typedef bool (HasVnetHdrLen)(NetClientState *, int);
>> +typedef bool (GetUsingVnetHdr)(NetClientState *);
>> typedef void (UsingVnetHdr)(NetClientState *, bool);
>> typedef void (SetOffload)(NetClientState *, int, int, int, int, int);
>> +typedef int (GetVnetHdrLen)(NetClientState *);
>> typedef void (SetVnetHdrLen)(NetClientState *, int);
>> typedef int (SetVnetLE)(NetClientState *, bool);
>> typedef int (SetVnetBE)(NetClientState *, bool);
>> @@ -84,8 +86,10 @@ typedef struct NetClientInfo {
>> HasUfo *has_ufo;
>> HasVnetHdr *has_vnet_hdr;
>> HasVnetHdrLen *has_vnet_hdr_len;
>> + GetUsingVnetHdr *get_using_vnet_hdr;
>> UsingVnetHdr *using_vnet_hdr;
>> SetOffload *set_offload;
>> + GetVnetHdrLen *get_vnet_hdr_len;
>> SetVnetHdrLen *set_vnet_hdr_len;
>> SetVnetLE *set_vnet_le;
>> SetVnetBE *set_vnet_be;
>> @@ -183,9 +187,11 @@ void qemu_format_nic_info_str(NetClientState *nc, uint8_t macaddr[6]);
>> bool qemu_has_ufo(NetClientState *nc);
>> bool qemu_has_vnet_hdr(NetClientState *nc);
>> bool qemu_has_vnet_hdr_len(NetClientState *nc, int len);
>> +bool qemu_get_using_vnet_hdr(NetClientState *nc);
>> void qemu_using_vnet_hdr(NetClientState *nc, bool enable);
>> void qemu_set_offload(NetClientState *nc, int csum, int tso4, int tso6,
>> int ecn, int ufo);
>> +int qemu_get_vnet_hdr_len(NetClientState *nc);
>> void qemu_set_vnet_hdr_len(NetClientState *nc, int len);
>> int qemu_set_vnet_le(NetClientState *nc, bool is_le);
>> int qemu_set_vnet_be(NetClientState *nc, bool is_be);
>> diff --git a/net/dump.c b/net/dump.c
>> index 6a63b15359..7d05f16ca7 100644
>> --- a/net/dump.c
>> +++ b/net/dump.c
>> @@ -61,12 +61,13 @@ struct pcap_sf_pkthdr {
>> uint32_t len;
>> };
>>
>> -static ssize_t dump_receive_iov(DumpState *s, const struct iovec *iov, int cnt)
>> +static ssize_t dump_receive_iov(DumpState *s, const struct iovec *iov, int cnt,
>> + int offset)
>> {
>> struct pcap_sf_pkthdr hdr;
>> int64_t ts;
>> int caplen;
>> - size_t size = iov_size(iov, cnt);
>> + size_t size = iov_size(iov, cnt) - offset;
>> struct iovec dumpiov[cnt + 1];
>>
>> /* Early return in case of previous error. */
>> @@ -84,7 +85,7 @@ static ssize_t dump_receive_iov(DumpState *s, const struct iovec *iov, int cnt)
>>
>> dumpiov[0].iov_base = &hdr;
>> dumpiov[0].iov_len = sizeof(hdr);
>> - cnt = iov_copy(&dumpiov[1], cnt, iov, cnt, 0, caplen);
>> + cnt = iov_copy(&dumpiov[1], cnt, iov, cnt, offset, caplen);
>>
>> if (writev(s->fd, dumpiov, cnt + 1) != sizeof(hdr) + caplen) {
>> error_report("network dump write error - stopping dump");
>> @@ -153,8 +154,10 @@ static ssize_t filter_dump_receive_iov(NetFilterState *nf, NetClientState *sndr,
>> int iovcnt, NetPacketSent *sent_cb)
>> {
>> NetFilterDumpState *nfds = FILTER_DUMP(nf);
>> + int offset = qemu_get_using_vnet_hdr(nf->netdev) ?
>> + qemu_get_vnet_hdr_len(nf->netdev) : 0;
>>
>> - dump_receive_iov(&nfds->ds, iov, iovcnt);
>> + dump_receive_iov(&nfds->ds, iov, iovcnt, offset);
>> return 0;
>> }
>>
>> diff --git a/net/net.c b/net/net.c
>> index 2d01472998..03f17de5fc 100644
>> --- a/net/net.c
>> +++ b/net/net.c
>> @@ -513,6 +513,15 @@ bool qemu_has_vnet_hdr_len(NetClientState *nc, int len)
>> return nc->info->has_vnet_hdr_len(nc, len);
>> }
>>
>> +bool qemu_get_using_vnet_hdr(NetClientState *nc)
>> +{
>> + if (!nc || !nc->info->get_using_vnet_hdr) {
>> + return false;
>> + }
>> +
>> + return nc->info->get_using_vnet_hdr(nc);
>> +}
>> +
>> void qemu_using_vnet_hdr(NetClientState *nc, bool enable)
>> {
>> if (!nc || !nc->info->using_vnet_hdr) {
>> @@ -532,6 +541,15 @@ void qemu_set_offload(NetClientState *nc, int csum, int tso4, int tso6,
>> nc->info->set_offload(nc, csum, tso4, tso6, ecn, ufo);
>> }
>>
>> +int qemu_get_vnet_hdr_len(NetClientState *nc)
>> +{
>> + if (!nc || !nc->info->get_vnet_hdr_len) {
>> + return 0;
>> + }
>> +
>> + return nc->info->get_vnet_hdr_len(nc);
>> +}
>> +
>> void qemu_set_vnet_hdr_len(NetClientState *nc, int len)
>> {
>> if (!nc || !nc->info->set_vnet_hdr_len) {
>> diff --git a/net/tap.c b/net/tap.c
>> index 7d7bc1dc5f..1bf085d422 100644
>> --- a/net/tap.c
>> +++ b/net/tap.c
>> @@ -255,6 +255,13 @@ static bool tap_has_vnet_hdr_len(NetClientState *nc, int len)
>> return !!tap_probe_vnet_hdr_len(s->fd, len);
>> }
>>
>> +static int tap_get_vnet_hdr_len(NetClientState *nc)
>> +{
>> + TAPState *s = DO_UPCAST(TAPState, nc, nc);
>> +
>> + return s->host_vnet_hdr_len;
>> +}
>> +
>> static void tap_set_vnet_hdr_len(NetClientState *nc, int len)
>> {
>> TAPState *s = DO_UPCAST(TAPState, nc, nc);
>> @@ -268,6 +275,13 @@ static void tap_set_vnet_hdr_len(NetClientState *nc, int len)
>> s->host_vnet_hdr_len = len;
>> }
>>
>> +static bool tap_get_using_vnet_hdr(NetClientState *nc)
>> +{
>> + TAPState *s = DO_UPCAST(TAPState, nc, nc);
>> +
>> + return s->using_vnet_hdr;
>> +}
>> +
>> static void tap_using_vnet_hdr(NetClientState *nc, bool using_vnet_hdr)
>> {
>> TAPState *s = DO_UPCAST(TAPState, nc, nc);
>> @@ -372,8 +386,10 @@ static NetClientInfo net_tap_info = {
>> .has_ufo = tap_has_ufo,
>> .has_vnet_hdr = tap_has_vnet_hdr,
>> .has_vnet_hdr_len = tap_has_vnet_hdr_len,
>> + .get_using_vnet_hdr = tap_get_using_vnet_hdr,
>> .using_vnet_hdr = tap_using_vnet_hdr,
>> .set_offload = tap_set_offload,
>> + .get_vnet_hdr_len = tap_get_vnet_hdr_len,
>> .set_vnet_hdr_len = tap_set_vnet_hdr_len,
>> .set_vnet_le = tap_set_vnet_le,
>> .set_vnet_be = tap_set_vnet_be,
>> --
>> 2.39.1
>
^ permalink raw reply [flat|nested] 33+ messages in thread* Re: [PATCH v4 20/28] net: Strip virtio-net header when dumping
2023-01-30 15:36 ` Akihiko Odaki
@ 2023-01-30 15:47 ` Michael S. Tsirkin
2023-01-31 2:36 ` Akihiko Odaki
0 siblings, 1 reply; 33+ messages in thread
From: Michael S. Tsirkin @ 2023-01-30 15:47 UTC (permalink / raw)
To: Akihiko Odaki
Cc: Jason Wang, Dmitry Fleytman, Thomas Huth, Laurent Vivier,
Paolo Bonzini, qemu-devel, qemu-ppc, Yan Vugenfirer,
Yuri Benditovich, Sriram Yagnaraman, Alexander Bulekov
On Tue, Jan 31, 2023 at 12:36:38AM +0900, Akihiko Odaki wrote:
> On 2023/01/31 0:12, Michael S. Tsirkin wrote:
> > On Mon, Jan 30, 2023 at 10:47:07PM +0900, Akihiko Odaki wrote:
> > > filter-dump specifiees Ethernet as PCAP LinkType, which does not expect
> > > virtio-net header. Having virtio-net header in such PCAP file breaks
> > > PCAP unconsumable. Unfortunately currently there is no LinkType for
> > > virtio-net so for now strip virtio-net header to convert the output to
> > > Ethernet.
> > >
> > > Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
> >
> > Probably means you need to calculate checksums and split gso too right?
>
> It was not necessary in my case as I used Wireshark and it tolerates wrong
> checksums and large packets (it even says "Checksum incorrect [maybe caused
> by 'TCP checksum offload'?]"). It was even more helpful to have raw packets
> instead of transformed packets for debugging purposes. Perhaps an option to
> transform packets may be added later if a need arises.
I think we should add LINKTYPE_VIRTIO. Very useful to debug a host of
checksum/segmentation issues. Want to hack it up?
> >
> > > ---
> > > include/net/net.h | 6 ++++++
> > > net/dump.c | 11 +++++++----
> > > net/net.c | 18 ++++++++++++++++++
> > > net/tap.c | 16 ++++++++++++++++
> > > 4 files changed, 47 insertions(+), 4 deletions(-)
> > >
> > > diff --git a/include/net/net.h b/include/net/net.h
> > > index dc20b31e9f..4b2d72b3fc 100644
> > > --- a/include/net/net.h
> > > +++ b/include/net/net.h
> > > @@ -56,8 +56,10 @@ typedef RxFilterInfo *(QueryRxFilter)(NetClientState *);
> > > typedef bool (HasUfo)(NetClientState *);
> > > typedef bool (HasVnetHdr)(NetClientState *);
> > > typedef bool (HasVnetHdrLen)(NetClientState *, int);
> > > +typedef bool (GetUsingVnetHdr)(NetClientState *);
> > > typedef void (UsingVnetHdr)(NetClientState *, bool);
> > > typedef void (SetOffload)(NetClientState *, int, int, int, int, int);
> > > +typedef int (GetVnetHdrLen)(NetClientState *);
> > > typedef void (SetVnetHdrLen)(NetClientState *, int);
> > > typedef int (SetVnetLE)(NetClientState *, bool);
> > > typedef int (SetVnetBE)(NetClientState *, bool);
> > > @@ -84,8 +86,10 @@ typedef struct NetClientInfo {
> > > HasUfo *has_ufo;
> > > HasVnetHdr *has_vnet_hdr;
> > > HasVnetHdrLen *has_vnet_hdr_len;
> > > + GetUsingVnetHdr *get_using_vnet_hdr;
> > > UsingVnetHdr *using_vnet_hdr;
> > > SetOffload *set_offload;
> > > + GetVnetHdrLen *get_vnet_hdr_len;
> > > SetVnetHdrLen *set_vnet_hdr_len;
> > > SetVnetLE *set_vnet_le;
> > > SetVnetBE *set_vnet_be;
> > > @@ -183,9 +187,11 @@ void qemu_format_nic_info_str(NetClientState *nc, uint8_t macaddr[6]);
> > > bool qemu_has_ufo(NetClientState *nc);
> > > bool qemu_has_vnet_hdr(NetClientState *nc);
> > > bool qemu_has_vnet_hdr_len(NetClientState *nc, int len);
> > > +bool qemu_get_using_vnet_hdr(NetClientState *nc);
> > > void qemu_using_vnet_hdr(NetClientState *nc, bool enable);
> > > void qemu_set_offload(NetClientState *nc, int csum, int tso4, int tso6,
> > > int ecn, int ufo);
> > > +int qemu_get_vnet_hdr_len(NetClientState *nc);
> > > void qemu_set_vnet_hdr_len(NetClientState *nc, int len);
> > > int qemu_set_vnet_le(NetClientState *nc, bool is_le);
> > > int qemu_set_vnet_be(NetClientState *nc, bool is_be);
> > > diff --git a/net/dump.c b/net/dump.c
> > > index 6a63b15359..7d05f16ca7 100644
> > > --- a/net/dump.c
> > > +++ b/net/dump.c
> > > @@ -61,12 +61,13 @@ struct pcap_sf_pkthdr {
> > > uint32_t len;
> > > };
> > > -static ssize_t dump_receive_iov(DumpState *s, const struct iovec *iov, int cnt)
> > > +static ssize_t dump_receive_iov(DumpState *s, const struct iovec *iov, int cnt,
> > > + int offset)
> > > {
> > > struct pcap_sf_pkthdr hdr;
> > > int64_t ts;
> > > int caplen;
> > > - size_t size = iov_size(iov, cnt);
> > > + size_t size = iov_size(iov, cnt) - offset;
> > > struct iovec dumpiov[cnt + 1];
> > > /* Early return in case of previous error. */
> > > @@ -84,7 +85,7 @@ static ssize_t dump_receive_iov(DumpState *s, const struct iovec *iov, int cnt)
> > > dumpiov[0].iov_base = &hdr;
> > > dumpiov[0].iov_len = sizeof(hdr);
> > > - cnt = iov_copy(&dumpiov[1], cnt, iov, cnt, 0, caplen);
> > > + cnt = iov_copy(&dumpiov[1], cnt, iov, cnt, offset, caplen);
> > > if (writev(s->fd, dumpiov, cnt + 1) != sizeof(hdr) + caplen) {
> > > error_report("network dump write error - stopping dump");
> > > @@ -153,8 +154,10 @@ static ssize_t filter_dump_receive_iov(NetFilterState *nf, NetClientState *sndr,
> > > int iovcnt, NetPacketSent *sent_cb)
> > > {
> > > NetFilterDumpState *nfds = FILTER_DUMP(nf);
> > > + int offset = qemu_get_using_vnet_hdr(nf->netdev) ?
> > > + qemu_get_vnet_hdr_len(nf->netdev) : 0;
> > > - dump_receive_iov(&nfds->ds, iov, iovcnt);
> > > + dump_receive_iov(&nfds->ds, iov, iovcnt, offset);
> > > return 0;
> > > }
> > > diff --git a/net/net.c b/net/net.c
> > > index 2d01472998..03f17de5fc 100644
> > > --- a/net/net.c
> > > +++ b/net/net.c
> > > @@ -513,6 +513,15 @@ bool qemu_has_vnet_hdr_len(NetClientState *nc, int len)
> > > return nc->info->has_vnet_hdr_len(nc, len);
> > > }
> > > +bool qemu_get_using_vnet_hdr(NetClientState *nc)
> > > +{
> > > + if (!nc || !nc->info->get_using_vnet_hdr) {
> > > + return false;
> > > + }
> > > +
> > > + return nc->info->get_using_vnet_hdr(nc);
> > > +}
> > > +
> > > void qemu_using_vnet_hdr(NetClientState *nc, bool enable)
> > > {
> > > if (!nc || !nc->info->using_vnet_hdr) {
> > > @@ -532,6 +541,15 @@ void qemu_set_offload(NetClientState *nc, int csum, int tso4, int tso6,
> > > nc->info->set_offload(nc, csum, tso4, tso6, ecn, ufo);
> > > }
> > > +int qemu_get_vnet_hdr_len(NetClientState *nc)
> > > +{
> > > + if (!nc || !nc->info->get_vnet_hdr_len) {
> > > + return 0;
> > > + }
> > > +
> > > + return nc->info->get_vnet_hdr_len(nc);
> > > +}
> > > +
> > > void qemu_set_vnet_hdr_len(NetClientState *nc, int len)
> > > {
> > > if (!nc || !nc->info->set_vnet_hdr_len) {
> > > diff --git a/net/tap.c b/net/tap.c
> > > index 7d7bc1dc5f..1bf085d422 100644
> > > --- a/net/tap.c
> > > +++ b/net/tap.c
> > > @@ -255,6 +255,13 @@ static bool tap_has_vnet_hdr_len(NetClientState *nc, int len)
> > > return !!tap_probe_vnet_hdr_len(s->fd, len);
> > > }
> > > +static int tap_get_vnet_hdr_len(NetClientState *nc)
> > > +{
> > > + TAPState *s = DO_UPCAST(TAPState, nc, nc);
> > > +
> > > + return s->host_vnet_hdr_len;
> > > +}
> > > +
> > > static void tap_set_vnet_hdr_len(NetClientState *nc, int len)
> > > {
> > > TAPState *s = DO_UPCAST(TAPState, nc, nc);
> > > @@ -268,6 +275,13 @@ static void tap_set_vnet_hdr_len(NetClientState *nc, int len)
> > > s->host_vnet_hdr_len = len;
> > > }
> > > +static bool tap_get_using_vnet_hdr(NetClientState *nc)
> > > +{
> > > + TAPState *s = DO_UPCAST(TAPState, nc, nc);
> > > +
> > > + return s->using_vnet_hdr;
> > > +}
> > > +
> > > static void tap_using_vnet_hdr(NetClientState *nc, bool using_vnet_hdr)
> > > {
> > > TAPState *s = DO_UPCAST(TAPState, nc, nc);
> > > @@ -372,8 +386,10 @@ static NetClientInfo net_tap_info = {
> > > .has_ufo = tap_has_ufo,
> > > .has_vnet_hdr = tap_has_vnet_hdr,
> > > .has_vnet_hdr_len = tap_has_vnet_hdr_len,
> > > + .get_using_vnet_hdr = tap_get_using_vnet_hdr,
> > > .using_vnet_hdr = tap_using_vnet_hdr,
> > > .set_offload = tap_set_offload,
> > > + .get_vnet_hdr_len = tap_get_vnet_hdr_len,
> > > .set_vnet_hdr_len = tap_set_vnet_hdr_len,
> > > .set_vnet_le = tap_set_vnet_le,
> > > .set_vnet_be = tap_set_vnet_be,
> > > --
> > > 2.39.1
> >
^ permalink raw reply [flat|nested] 33+ messages in thread* Re: [PATCH v4 20/28] net: Strip virtio-net header when dumping
2023-01-30 15:47 ` Michael S. Tsirkin
@ 2023-01-31 2:36 ` Akihiko Odaki
0 siblings, 0 replies; 33+ messages in thread
From: Akihiko Odaki @ 2023-01-31 2:36 UTC (permalink / raw)
To: Michael S. Tsirkin
Cc: Jason Wang, Dmitry Fleytman, Thomas Huth, Laurent Vivier,
Paolo Bonzini, qemu-devel, qemu-ppc, Yan Vugenfirer,
Yuri Benditovich, Sriram Yagnaraman, Alexander Bulekov
On 2023/01/31 0:47, Michael S. Tsirkin wrote:
> On Tue, Jan 31, 2023 at 12:36:38AM +0900, Akihiko Odaki wrote:
>> On 2023/01/31 0:12, Michael S. Tsirkin wrote:
>>> On Mon, Jan 30, 2023 at 10:47:07PM +0900, Akihiko Odaki wrote:
>>>> filter-dump specifiees Ethernet as PCAP LinkType, which does not expect
>>>> virtio-net header. Having virtio-net header in such PCAP file breaks
>>>> PCAP unconsumable. Unfortunately currently there is no LinkType for
>>>> virtio-net so for now strip virtio-net header to convert the output to
>>>> Ethernet.
>>>>
>>>> Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
>>>
>>> Probably means you need to calculate checksums and split gso too right?
>>
>> It was not necessary in my case as I used Wireshark and it tolerates wrong
>> checksums and large packets (it even says "Checksum incorrect [maybe caused
>> by 'TCP checksum offload'?]"). It was even more helpful to have raw packets
>> instead of transformed packets for debugging purposes. Perhaps an option to
>> transform packets may be added later if a need arises.
>
> I think we should add LINKTYPE_VIRTIO. Very useful to debug a host of
> checksum/segmentation issues. Want to hack it up?
I'd rather like to land this patch as is. I think patching Wireshark so
that it repsects virtio-net flags is a non-trivial task. Even if
Wireshark is patched, it takes time until the patched version becomes
available, and other tooling may not work with the new linktype.
In my opinion, virtio-net header is not worth much when debugging as it
only contains some simple flags and is unlikely to be corrupted. The
logic to determine whether the flags should be set is also simple (e.g.,
if TCP segmentation offload is enabled and it is TCP/IPv4, set
VIRTIO_NET_HDR_GSO_TCPV4).
>
>>>
>>>> ---
>>>> include/net/net.h | 6 ++++++
>>>> net/dump.c | 11 +++++++----
>>>> net/net.c | 18 ++++++++++++++++++
>>>> net/tap.c | 16 ++++++++++++++++
>>>> 4 files changed, 47 insertions(+), 4 deletions(-)
>>>>
>>>> diff --git a/include/net/net.h b/include/net/net.h
>>>> index dc20b31e9f..4b2d72b3fc 100644
>>>> --- a/include/net/net.h
>>>> +++ b/include/net/net.h
>>>> @@ -56,8 +56,10 @@ typedef RxFilterInfo *(QueryRxFilter)(NetClientState *);
>>>> typedef bool (HasUfo)(NetClientState *);
>>>> typedef bool (HasVnetHdr)(NetClientState *);
>>>> typedef bool (HasVnetHdrLen)(NetClientState *, int);
>>>> +typedef bool (GetUsingVnetHdr)(NetClientState *);
>>>> typedef void (UsingVnetHdr)(NetClientState *, bool);
>>>> typedef void (SetOffload)(NetClientState *, int, int, int, int, int);
>>>> +typedef int (GetVnetHdrLen)(NetClientState *);
>>>> typedef void (SetVnetHdrLen)(NetClientState *, int);
>>>> typedef int (SetVnetLE)(NetClientState *, bool);
>>>> typedef int (SetVnetBE)(NetClientState *, bool);
>>>> @@ -84,8 +86,10 @@ typedef struct NetClientInfo {
>>>> HasUfo *has_ufo;
>>>> HasVnetHdr *has_vnet_hdr;
>>>> HasVnetHdrLen *has_vnet_hdr_len;
>>>> + GetUsingVnetHdr *get_using_vnet_hdr;
>>>> UsingVnetHdr *using_vnet_hdr;
>>>> SetOffload *set_offload;
>>>> + GetVnetHdrLen *get_vnet_hdr_len;
>>>> SetVnetHdrLen *set_vnet_hdr_len;
>>>> SetVnetLE *set_vnet_le;
>>>> SetVnetBE *set_vnet_be;
>>>> @@ -183,9 +187,11 @@ void qemu_format_nic_info_str(NetClientState *nc, uint8_t macaddr[6]);
>>>> bool qemu_has_ufo(NetClientState *nc);
>>>> bool qemu_has_vnet_hdr(NetClientState *nc);
>>>> bool qemu_has_vnet_hdr_len(NetClientState *nc, int len);
>>>> +bool qemu_get_using_vnet_hdr(NetClientState *nc);
>>>> void qemu_using_vnet_hdr(NetClientState *nc, bool enable);
>>>> void qemu_set_offload(NetClientState *nc, int csum, int tso4, int tso6,
>>>> int ecn, int ufo);
>>>> +int qemu_get_vnet_hdr_len(NetClientState *nc);
>>>> void qemu_set_vnet_hdr_len(NetClientState *nc, int len);
>>>> int qemu_set_vnet_le(NetClientState *nc, bool is_le);
>>>> int qemu_set_vnet_be(NetClientState *nc, bool is_be);
>>>> diff --git a/net/dump.c b/net/dump.c
>>>> index 6a63b15359..7d05f16ca7 100644
>>>> --- a/net/dump.c
>>>> +++ b/net/dump.c
>>>> @@ -61,12 +61,13 @@ struct pcap_sf_pkthdr {
>>>> uint32_t len;
>>>> };
>>>> -static ssize_t dump_receive_iov(DumpState *s, const struct iovec *iov, int cnt)
>>>> +static ssize_t dump_receive_iov(DumpState *s, const struct iovec *iov, int cnt,
>>>> + int offset)
>>>> {
>>>> struct pcap_sf_pkthdr hdr;
>>>> int64_t ts;
>>>> int caplen;
>>>> - size_t size = iov_size(iov, cnt);
>>>> + size_t size = iov_size(iov, cnt) - offset;
>>>> struct iovec dumpiov[cnt + 1];
>>>> /* Early return in case of previous error. */
>>>> @@ -84,7 +85,7 @@ static ssize_t dump_receive_iov(DumpState *s, const struct iovec *iov, int cnt)
>>>> dumpiov[0].iov_base = &hdr;
>>>> dumpiov[0].iov_len = sizeof(hdr);
>>>> - cnt = iov_copy(&dumpiov[1], cnt, iov, cnt, 0, caplen);
>>>> + cnt = iov_copy(&dumpiov[1], cnt, iov, cnt, offset, caplen);
>>>> if (writev(s->fd, dumpiov, cnt + 1) != sizeof(hdr) + caplen) {
>>>> error_report("network dump write error - stopping dump");
>>>> @@ -153,8 +154,10 @@ static ssize_t filter_dump_receive_iov(NetFilterState *nf, NetClientState *sndr,
>>>> int iovcnt, NetPacketSent *sent_cb)
>>>> {
>>>> NetFilterDumpState *nfds = FILTER_DUMP(nf);
>>>> + int offset = qemu_get_using_vnet_hdr(nf->netdev) ?
>>>> + qemu_get_vnet_hdr_len(nf->netdev) : 0;
>>>> - dump_receive_iov(&nfds->ds, iov, iovcnt);
>>>> + dump_receive_iov(&nfds->ds, iov, iovcnt, offset);
>>>> return 0;
>>>> }
>>>> diff --git a/net/net.c b/net/net.c
>>>> index 2d01472998..03f17de5fc 100644
>>>> --- a/net/net.c
>>>> +++ b/net/net.c
>>>> @@ -513,6 +513,15 @@ bool qemu_has_vnet_hdr_len(NetClientState *nc, int len)
>>>> return nc->info->has_vnet_hdr_len(nc, len);
>>>> }
>>>> +bool qemu_get_using_vnet_hdr(NetClientState *nc)
>>>> +{
>>>> + if (!nc || !nc->info->get_using_vnet_hdr) {
>>>> + return false;
>>>> + }
>>>> +
>>>> + return nc->info->get_using_vnet_hdr(nc);
>>>> +}
>>>> +
>>>> void qemu_using_vnet_hdr(NetClientState *nc, bool enable)
>>>> {
>>>> if (!nc || !nc->info->using_vnet_hdr) {
>>>> @@ -532,6 +541,15 @@ void qemu_set_offload(NetClientState *nc, int csum, int tso4, int tso6,
>>>> nc->info->set_offload(nc, csum, tso4, tso6, ecn, ufo);
>>>> }
>>>> +int qemu_get_vnet_hdr_len(NetClientState *nc)
>>>> +{
>>>> + if (!nc || !nc->info->get_vnet_hdr_len) {
>>>> + return 0;
>>>> + }
>>>> +
>>>> + return nc->info->get_vnet_hdr_len(nc);
>>>> +}
>>>> +
>>>> void qemu_set_vnet_hdr_len(NetClientState *nc, int len)
>>>> {
>>>> if (!nc || !nc->info->set_vnet_hdr_len) {
>>>> diff --git a/net/tap.c b/net/tap.c
>>>> index 7d7bc1dc5f..1bf085d422 100644
>>>> --- a/net/tap.c
>>>> +++ b/net/tap.c
>>>> @@ -255,6 +255,13 @@ static bool tap_has_vnet_hdr_len(NetClientState *nc, int len)
>>>> return !!tap_probe_vnet_hdr_len(s->fd, len);
>>>> }
>>>> +static int tap_get_vnet_hdr_len(NetClientState *nc)
>>>> +{
>>>> + TAPState *s = DO_UPCAST(TAPState, nc, nc);
>>>> +
>>>> + return s->host_vnet_hdr_len;
>>>> +}
>>>> +
>>>> static void tap_set_vnet_hdr_len(NetClientState *nc, int len)
>>>> {
>>>> TAPState *s = DO_UPCAST(TAPState, nc, nc);
>>>> @@ -268,6 +275,13 @@ static void tap_set_vnet_hdr_len(NetClientState *nc, int len)
>>>> s->host_vnet_hdr_len = len;
>>>> }
>>>> +static bool tap_get_using_vnet_hdr(NetClientState *nc)
>>>> +{
>>>> + TAPState *s = DO_UPCAST(TAPState, nc, nc);
>>>> +
>>>> + return s->using_vnet_hdr;
>>>> +}
>>>> +
>>>> static void tap_using_vnet_hdr(NetClientState *nc, bool using_vnet_hdr)
>>>> {
>>>> TAPState *s = DO_UPCAST(TAPState, nc, nc);
>>>> @@ -372,8 +386,10 @@ static NetClientInfo net_tap_info = {
>>>> .has_ufo = tap_has_ufo,
>>>> .has_vnet_hdr = tap_has_vnet_hdr,
>>>> .has_vnet_hdr_len = tap_has_vnet_hdr_len,
>>>> + .get_using_vnet_hdr = tap_get_using_vnet_hdr,
>>>> .using_vnet_hdr = tap_using_vnet_hdr,
>>>> .set_offload = tap_set_offload,
>>>> + .get_vnet_hdr_len = tap_get_vnet_hdr_len,
>>>> .set_vnet_hdr_len = tap_set_vnet_hdr_len,
>>>> .set_vnet_le = tap_set_vnet_le,
>>>> .set_vnet_be = tap_set_vnet_be,
>>>> --
>>>> 2.39.1
>>>
>
^ permalink raw reply [flat|nested] 33+ messages in thread
* [PATCH v4 21/28] hw/net/net_tx_pkt: Automatically determine if virtio-net header is used
2023-01-30 13:46 [PATCH v4 00/28] e1000x cleanups (preliminary for IGB) Akihiko Odaki
` (19 preceding siblings ...)
2023-01-30 13:47 ` [PATCH v4 20/28] net: Strip virtio-net header when dumping Akihiko Odaki
@ 2023-01-30 13:47 ` Akihiko Odaki
2023-01-30 13:47 ` [PATCH v4 22/28] hw/net/net_rx_pkt: Remove net_rx_pkt_has_virt_hdr Akihiko Odaki
` (6 subsequent siblings)
27 siblings, 0 replies; 33+ messages in thread
From: Akihiko Odaki @ 2023-01-30 13:47 UTC (permalink / raw)
Cc: Jason Wang, Dmitry Fleytman, Thomas Huth, Laurent Vivier,
Paolo Bonzini, qemu-devel, qemu-ppc, Michael S . Tsirkin,
Yan Vugenfirer, Yuri Benditovich, Sriram Yagnaraman,
Alexander Bulekov, Akihiko Odaki
The new function qemu_get_using_vnet_hdr() allows to automatically
determine if virtio-net header is used.
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
---
hw/net/e1000e_core.c | 3 +--
hw/net/net_tx_pkt.c | 19 ++++++++++---------
hw/net/net_tx_pkt.h | 3 +--
hw/net/vmxnet3.c | 6 ++----
4 files changed, 14 insertions(+), 17 deletions(-)
diff --git a/hw/net/e1000e_core.c b/hw/net/e1000e_core.c
index 38d374fba3..954a007151 100644
--- a/hw/net/e1000e_core.c
+++ b/hw/net/e1000e_core.c
@@ -3376,8 +3376,7 @@ e1000e_core_pci_realize(E1000ECore *core,
qemu_add_vm_change_state_handler(e1000e_vm_state_change, core);
for (i = 0; i < E1000E_NUM_QUEUES; i++) {
- net_tx_pkt_init(&core->tx[i].tx_pkt, core->owner,
- E1000E_MAX_TX_FRAGS, core->has_vnet);
+ net_tx_pkt_init(&core->tx[i].tx_pkt, core->owner, E1000E_MAX_TX_FRAGS);
}
net_rx_pkt_init(&core->rx_pkt, core->has_vnet);
diff --git a/hw/net/net_tx_pkt.c b/hw/net/net_tx_pkt.c
index 8a23899a4d..cf46c8457f 100644
--- a/hw/net/net_tx_pkt.c
+++ b/hw/net/net_tx_pkt.c
@@ -35,7 +35,6 @@ struct NetTxPkt {
PCIDevice *pci_dev;
struct virtio_net_hdr virt_hdr;
- bool has_virt_hdr;
struct iovec *raw;
uint32_t raw_frags;
@@ -59,7 +58,7 @@ struct NetTxPkt {
};
void net_tx_pkt_init(struct NetTxPkt **pkt, PCIDevice *pci_dev,
- uint32_t max_frags, bool has_virt_hdr)
+ uint32_t max_frags)
{
struct NetTxPkt *p = g_malloc0(sizeof *p);
@@ -71,10 +70,8 @@ void net_tx_pkt_init(struct NetTxPkt **pkt, PCIDevice *pci_dev,
p->max_payload_frags = max_frags;
p->max_raw_frags = max_frags;
- p->has_virt_hdr = has_virt_hdr;
p->vec[NET_TX_PKT_VHDR_FRAG].iov_base = &p->virt_hdr;
- p->vec[NET_TX_PKT_VHDR_FRAG].iov_len =
- p->has_virt_hdr ? sizeof p->virt_hdr : 0;
+ p->vec[NET_TX_PKT_VHDR_FRAG].iov_len = sizeof p->virt_hdr;
p->vec[NET_TX_PKT_L2HDR_FRAG].iov_base = &p->l2_hdr;
p->vec[NET_TX_PKT_L3HDR_FRAG].iov_base = &p->l3_hdr;
@@ -617,9 +614,11 @@ static bool net_tx_pkt_do_sw_fragmentation(struct NetTxPkt *pkt,
bool net_tx_pkt_send(struct NetTxPkt *pkt, NetClientState *nc)
{
+ bool using_vnet_hdr = qemu_get_using_vnet_hdr(nc->peer);
+
assert(pkt);
- if (!pkt->has_virt_hdr &&
+ if (!using_vnet_hdr &&
pkt->virt_hdr.flags & VIRTIO_NET_HDR_F_NEEDS_CSUM) {
net_tx_pkt_do_sw_csum(pkt);
}
@@ -636,11 +635,13 @@ bool net_tx_pkt_send(struct NetTxPkt *pkt, NetClientState *nc)
}
}
- if (pkt->has_virt_hdr ||
+ if (using_vnet_hdr ||
pkt->virt_hdr.gso_type == VIRTIO_NET_HDR_GSO_NONE) {
+ int index = using_vnet_hdr ?
+ NET_TX_PKT_VHDR_FRAG : NET_TX_PKT_L2HDR_FRAG;
net_tx_pkt_fix_ip6_payload_len(pkt);
- net_tx_pkt_sendv(pkt, nc, pkt->vec,
- pkt->payload_frags + NET_TX_PKT_PL_START_FRAG);
+ net_tx_pkt_sendv(pkt, nc, pkt->vec + index,
+ pkt->payload_frags + NET_TX_PKT_PL_START_FRAG - index);
return true;
}
diff --git a/hw/net/net_tx_pkt.h b/hw/net/net_tx_pkt.h
index 2e38a5fa69..8d3faa42fb 100644
--- a/hw/net/net_tx_pkt.h
+++ b/hw/net/net_tx_pkt.h
@@ -32,10 +32,9 @@ struct NetTxPkt;
* @pkt: packet pointer
* @pci_dev: PCI device processing this packet
* @max_frags: max tx ip fragments
- * @has_virt_hdr: device uses virtio header.
*/
void net_tx_pkt_init(struct NetTxPkt **pkt, PCIDevice *pci_dev,
- uint32_t max_frags, bool has_virt_hdr);
+ uint32_t max_frags);
/**
* Clean all tx packet resources.
diff --git a/hw/net/vmxnet3.c b/hw/net/vmxnet3.c
index c63bbb59bd..8c3f5d6e14 100644
--- a/hw/net/vmxnet3.c
+++ b/hw/net/vmxnet3.c
@@ -1521,8 +1521,7 @@ static void vmxnet3_activate_device(VMXNET3State *s)
/* Preallocate TX packet wrapper */
VMW_CFPRN("Max TX fragments is %u", s->max_tx_frags);
- net_tx_pkt_init(&s->tx_pkt, PCI_DEVICE(s),
- s->max_tx_frags, s->peer_has_vhdr);
+ net_tx_pkt_init(&s->tx_pkt, PCI_DEVICE(s), s->max_tx_frags);
net_rx_pkt_init(&s->rx_pkt, s->peer_has_vhdr);
/* Read rings memory locations for RX queues */
@@ -2402,8 +2401,7 @@ static int vmxnet3_post_load(void *opaque, int version_id)
{
VMXNET3State *s = opaque;
- net_tx_pkt_init(&s->tx_pkt, PCI_DEVICE(s),
- s->max_tx_frags, s->peer_has_vhdr);
+ net_tx_pkt_init(&s->tx_pkt, PCI_DEVICE(s), s->max_tx_frags);
net_rx_pkt_init(&s->rx_pkt, s->peer_has_vhdr);
if (s->msix_used) {
--
2.39.1
^ permalink raw reply related [flat|nested] 33+ messages in thread* [PATCH v4 22/28] hw/net/net_rx_pkt: Remove net_rx_pkt_has_virt_hdr
2023-01-30 13:46 [PATCH v4 00/28] e1000x cleanups (preliminary for IGB) Akihiko Odaki
` (20 preceding siblings ...)
2023-01-30 13:47 ` [PATCH v4 21/28] hw/net/net_tx_pkt: Automatically determine if virtio-net header is used Akihiko Odaki
@ 2023-01-30 13:47 ` Akihiko Odaki
2023-01-30 13:47 ` [PATCH v4 23/28] e1000e: Perform software segmentation for loopback Akihiko Odaki
` (5 subsequent siblings)
27 siblings, 0 replies; 33+ messages in thread
From: Akihiko Odaki @ 2023-01-30 13:47 UTC (permalink / raw)
Cc: Jason Wang, Dmitry Fleytman, Thomas Huth, Laurent Vivier,
Paolo Bonzini, qemu-devel, qemu-ppc, Michael S . Tsirkin,
Yan Vugenfirer, Yuri Benditovich, Sriram Yagnaraman,
Alexander Bulekov, Akihiko Odaki
When virtio-net header is not set, net_rx_pkt_get_vhdr() returns
zero-filled virtio_net_hdr, which is actually valid. In fact, tap device
uses zero-filled virtio_net_hdr when virtio-net header is not provided
by the peer. Therefore, we can just remove net_rx_pkt_has_virt_hdr() and
always assume NetTxPkt has a valid virtio-net header.
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
---
hw/net/e1000e_core.c | 16 ++++------------
hw/net/net_rx_pkt.c | 11 +----------
hw/net/net_rx_pkt.h | 12 +-----------
hw/net/trace-events | 1 -
hw/net/virtio-net.c | 2 +-
hw/net/vmxnet3.c | 12 ++----------
6 files changed, 9 insertions(+), 45 deletions(-)
diff --git a/hw/net/e1000e_core.c b/hw/net/e1000e_core.c
index 954a007151..95245c42f5 100644
--- a/hw/net/e1000e_core.c
+++ b/hw/net/e1000e_core.c
@@ -1240,12 +1240,6 @@ e1000e_build_rx_metadata(E1000ECore *core,
goto func_exit;
}
- if (!net_rx_pkt_has_virt_hdr(pkt)) {
- trace_e1000e_rx_metadata_no_virthdr();
- e1000e_verify_csum_in_sw(core, pkt, status_flags, istcp, isudp);
- goto func_exit;
- }
-
vhdr = net_rx_pkt_get_vhdr(pkt);
if (!(vhdr->flags & VIRTIO_NET_HDR_F_DATA_VALID) &&
@@ -1671,12 +1665,10 @@ e1000e_write_packet_to_guest(E1000ECore *core, struct NetRxPkt *pkt,
static inline void
e1000e_rx_fix_l4_csum(E1000ECore *core, struct NetRxPkt *pkt)
{
- if (net_rx_pkt_has_virt_hdr(pkt)) {
- struct virtio_net_hdr *vhdr = net_rx_pkt_get_vhdr(pkt);
+ struct virtio_net_hdr *vhdr = net_rx_pkt_get_vhdr(pkt);
- if (vhdr->flags & VIRTIO_NET_HDR_F_NEEDS_CSUM) {
- net_rx_pkt_fix_l4_csum(pkt);
- }
+ if (vhdr->flags & VIRTIO_NET_HDR_F_NEEDS_CSUM) {
+ net_rx_pkt_fix_l4_csum(pkt);
}
}
@@ -3379,7 +3371,7 @@ e1000e_core_pci_realize(E1000ECore *core,
net_tx_pkt_init(&core->tx[i].tx_pkt, core->owner, E1000E_MAX_TX_FRAGS);
}
- net_rx_pkt_init(&core->rx_pkt, core->has_vnet);
+ net_rx_pkt_init(&core->rx_pkt);
e1000x_core_prepare_eeprom(core->eeprom,
eeprom_templ,
diff --git a/hw/net/net_rx_pkt.c b/hw/net/net_rx_pkt.c
index 1e1c504e42..b309c2f476 100644
--- a/hw/net/net_rx_pkt.c
+++ b/hw/net/net_rx_pkt.c
@@ -30,7 +30,6 @@ struct NetRxPkt {
uint32_t tot_len;
uint16_t tci;
size_t ehdr_buf_len;
- bool has_virt_hdr;
eth_pkt_types_e packet_type;
/* Analysis results */
@@ -48,10 +47,9 @@ struct NetRxPkt {
eth_l4_hdr_info l4hdr_info;
};
-void net_rx_pkt_init(struct NetRxPkt **pkt, bool has_virt_hdr)
+void net_rx_pkt_init(struct NetRxPkt **pkt)
{
struct NetRxPkt *p = g_malloc0(sizeof *p);
- p->has_virt_hdr = has_virt_hdr;
p->vec = NULL;
p->vec_len_total = 0;
*pkt = p;
@@ -472,13 +470,6 @@ bool net_rx_pkt_is_vlan_stripped(struct NetRxPkt *pkt)
return pkt->ehdr_buf_len ? true : false;
}
-bool net_rx_pkt_has_virt_hdr(struct NetRxPkt *pkt)
-{
- assert(pkt);
-
- return pkt->has_virt_hdr;
-}
-
uint16_t net_rx_pkt_get_vlan_tag(struct NetRxPkt *pkt)
{
assert(pkt);
diff --git a/hw/net/net_rx_pkt.h b/hw/net/net_rx_pkt.h
index 048e3461f0..7277907a22 100644
--- a/hw/net/net_rx_pkt.h
+++ b/hw/net/net_rx_pkt.h
@@ -37,10 +37,9 @@ void net_rx_pkt_uninit(struct NetRxPkt *pkt);
* Init function for rx packet functionality
*
* @pkt: packet pointer
- * @has_virt_hdr: device uses virtio header
*
*/
-void net_rx_pkt_init(struct NetRxPkt **pkt, bool has_virt_hdr);
+void net_rx_pkt_init(struct NetRxPkt **pkt);
/**
* returns total length of data attached to rx context
@@ -214,15 +213,6 @@ uint16_t net_rx_pkt_get_vlan_tag(struct NetRxPkt *pkt);
*/
bool net_rx_pkt_is_vlan_stripped(struct NetRxPkt *pkt);
-/**
- * notifies caller if the packet has virtio header
- *
- * @pkt: packet
- * @ret: true if packet has virtio header, false otherwize
- *
- */
-bool net_rx_pkt_has_virt_hdr(struct NetRxPkt *pkt);
-
/**
* attach scatter-gather data to rx packet
*
diff --git a/hw/net/trace-events b/hw/net/trace-events
index c98ad12537..f7257a0693 100644
--- a/hw/net/trace-events
+++ b/hw/net/trace-events
@@ -188,7 +188,6 @@ e1000e_rx_metadata_rss(uint32_t rss, uint32_t mrq) "RSS data: rss: 0x%X, mrq: 0x
e1000e_rx_metadata_ip_id(uint16_t ip_id) "the IPv4 ID is 0x%X"
e1000e_rx_metadata_ack(void) "the packet is TCP ACK"
e1000e_rx_metadata_pkt_type(uint32_t pkt_type) "the packet type is %u"
-e1000e_rx_metadata_no_virthdr(void) "the packet has no virt-header"
e1000e_rx_metadata_virthdr_no_csum_info(void) "virt-header does not contain checksum info"
e1000e_rx_metadata_l3_cso_disabled(void) "IP4 CSO is disabled"
e1000e_rx_metadata_l4_cso_disabled(void) "TCP/UDP CSO is disabled"
diff --git a/hw/net/virtio-net.c b/hw/net/virtio-net.c
index 3ae909041a..1795e1aa7d 100644
--- a/hw/net/virtio-net.c
+++ b/hw/net/virtio-net.c
@@ -3703,7 +3703,7 @@ static void virtio_net_device_realize(DeviceState *dev, Error **errp)
QTAILQ_INIT(&n->rsc_chains);
n->qdev = dev;
- net_rx_pkt_init(&n->rx_pkt, false);
+ net_rx_pkt_init(&n->rx_pkt);
if (virtio_has_feature(n->host_features, VIRTIO_NET_F_RSS)) {
virtio_net_load_ebpf(n);
diff --git a/hw/net/vmxnet3.c b/hw/net/vmxnet3.c
index 8c3f5d6e14..f0ded333aa 100644
--- a/hw/net/vmxnet3.c
+++ b/hw/net/vmxnet3.c
@@ -851,10 +851,6 @@ static void vmxnet3_rx_need_csum_calculate(struct NetRxPkt *pkt,
uint8_t *data;
int len;
- if (!net_rx_pkt_has_virt_hdr(pkt)) {
- return;
- }
-
vhdr = net_rx_pkt_get_vhdr(pkt);
if (!VMXNET_FLAG_IS_SET(vhdr->flags, VIRTIO_NET_HDR_F_NEEDS_CSUM)) {
return;
@@ -898,10 +894,6 @@ static void vmxnet3_rx_update_descr(struct NetRxPkt *pkt,
rxcd->tci = net_rx_pkt_get_vlan_tag(pkt);
}
- if (!net_rx_pkt_has_virt_hdr(pkt)) {
- goto nocsum;
- }
-
vhdr = net_rx_pkt_get_vhdr(pkt);
/*
* Checksum is valid when lower level tell so or when lower level
@@ -1522,7 +1514,7 @@ static void vmxnet3_activate_device(VMXNET3State *s)
/* Preallocate TX packet wrapper */
VMW_CFPRN("Max TX fragments is %u", s->max_tx_frags);
net_tx_pkt_init(&s->tx_pkt, PCI_DEVICE(s), s->max_tx_frags);
- net_rx_pkt_init(&s->rx_pkt, s->peer_has_vhdr);
+ net_rx_pkt_init(&s->rx_pkt);
/* Read rings memory locations for RX queues */
for (i = 0; i < s->rxq_num; i++) {
@@ -2402,7 +2394,7 @@ static int vmxnet3_post_load(void *opaque, int version_id)
VMXNET3State *s = opaque;
net_tx_pkt_init(&s->tx_pkt, PCI_DEVICE(s), s->max_tx_frags);
- net_rx_pkt_init(&s->rx_pkt, s->peer_has_vhdr);
+ net_rx_pkt_init(&s->rx_pkt);
if (s->msix_used) {
vmxnet3_use_msix_vectors(s, VMXNET3_MAX_INTRS);
--
2.39.1
^ permalink raw reply related [flat|nested] 33+ messages in thread* [PATCH v4 23/28] e1000e: Perform software segmentation for loopback
2023-01-30 13:46 [PATCH v4 00/28] e1000x cleanups (preliminary for IGB) Akihiko Odaki
` (21 preceding siblings ...)
2023-01-30 13:47 ` [PATCH v4 22/28] hw/net/net_rx_pkt: Remove net_rx_pkt_has_virt_hdr Akihiko Odaki
@ 2023-01-30 13:47 ` Akihiko Odaki
2023-01-30 13:47 ` [PATCH v4 24/28] hw/net/net_tx_pkt: Implement TCP segmentation Akihiko Odaki
` (4 subsequent siblings)
27 siblings, 0 replies; 33+ messages in thread
From: Akihiko Odaki @ 2023-01-30 13:47 UTC (permalink / raw)
Cc: Jason Wang, Dmitry Fleytman, Thomas Huth, Laurent Vivier,
Paolo Bonzini, qemu-devel, qemu-ppc, Michael S . Tsirkin,
Yan Vugenfirer, Yuri Benditovich, Sriram Yagnaraman,
Alexander Bulekov, Akihiko Odaki
e1000e didn't perform software segmentation for loopback if virtio-net
header is enabled, which is wrong.
To fix the problem, introduce net_tx_pkt_send_custom(), which allows the
caller to specify whether offloading should be assumed or not.
net_tx_pkt_send_custom() also allows the caller to provide a custom
sending function. Packets with virtio-net headers and ones without
virtio-net headers will be provided at the same time so the function
can choose the preferred version. In case of e1000e loopback, it prefers
to have virtio-net headers as they allows to skip the checksum
verification if VIRTIO_NET_HDR_F_DATA_VALID is set.
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
---
hw/net/e1000e_core.c | 27 ++++++++++++++--
hw/net/net_rx_pkt.c | 7 ++++
hw/net/net_rx_pkt.h | 8 +++++
hw/net/net_tx_pkt.c | 76 +++++++++++++++++++++-----------------------
hw/net/net_tx_pkt.h | 21 ++++++------
5 files changed, 88 insertions(+), 51 deletions(-)
diff --git a/hw/net/e1000e_core.c b/hw/net/e1000e_core.c
index 95245c42f5..ff93547f88 100644
--- a/hw/net/e1000e_core.c
+++ b/hw/net/e1000e_core.c
@@ -61,6 +61,10 @@ union e1000_rx_desc_union {
union e1000_rx_desc_packet_split packet_split;
};
+static ssize_t
+e1000e_receive_internal(E1000ECore *core, const struct iovec *iov, int iovcnt,
+ bool has_vnet);
+
static inline void
e1000e_set_interrupt_cause(E1000ECore *core, uint32_t val);
@@ -655,6 +659,15 @@ e1000e_setup_tx_offloads(E1000ECore *core, struct e1000e_tx *tx)
return true;
}
+static void e1000e_tx_pkt_callback(void *core,
+ const struct iovec *iov,
+ int iovcnt,
+ const struct iovec *virt_iov,
+ int virt_iovcnt)
+{
+ e1000e_receive_internal(core, virt_iov, virt_iovcnt, true);
+}
+
static bool
e1000e_tx_pkt_send(E1000ECore *core, struct e1000e_tx *tx, int queue_index)
{
@@ -669,7 +682,8 @@ e1000e_tx_pkt_send(E1000ECore *core, struct e1000e_tx *tx, int queue_index)
if ((core->phy[0][MII_BMCR] & MII_BMCR_LOOPBACK) ||
((core->mac[RCTL] & E1000_RCTL_LBM_MAC) == E1000_RCTL_LBM_MAC)) {
- return net_tx_pkt_send_loopback(tx->tx_pkt, queue);
+ return net_tx_pkt_send_custom(tx->tx_pkt, false,
+ e1000e_tx_pkt_callback, core);
} else {
return net_tx_pkt_send(tx->tx_pkt, queue);
}
@@ -1674,6 +1688,13 @@ e1000e_rx_fix_l4_csum(E1000ECore *core, struct NetRxPkt *pkt)
ssize_t
e1000e_receive_iov(E1000ECore *core, const struct iovec *iov, int iovcnt)
+{
+ return e1000e_receive_internal(core, iov, iovcnt, core->has_vnet);
+}
+
+static ssize_t
+e1000e_receive_internal(E1000ECore *core, const struct iovec *iov, int iovcnt,
+ bool has_vnet)
{
static const int maximum_ethernet_hdr_len = (ETH_HLEN + 4);
@@ -1696,9 +1717,11 @@ e1000e_receive_iov(E1000ECore *core, const struct iovec *iov, int iovcnt)
}
/* Pull virtio header in */
- if (core->has_vnet) {
+ if (has_vnet) {
net_rx_pkt_set_vhdr_iovec(core->rx_pkt, iov, iovcnt);
iov_ofs = sizeof(struct virtio_net_hdr);
+ } else {
+ net_rx_pkt_unset_vhdr(core->rx_pkt);
}
filter_buf = iov->iov_base + iov_ofs;
diff --git a/hw/net/net_rx_pkt.c b/hw/net/net_rx_pkt.c
index b309c2f476..a53e7561c5 100644
--- a/hw/net/net_rx_pkt.c
+++ b/hw/net/net_rx_pkt.c
@@ -463,6 +463,13 @@ void net_rx_pkt_set_vhdr_iovec(struct NetRxPkt *pkt,
iov_to_buf(iov, iovcnt, 0, &pkt->virt_hdr, sizeof pkt->virt_hdr);
}
+void net_rx_pkt_unset_vhdr(struct NetRxPkt *pkt)
+{
+ assert(pkt);
+
+ memset(&pkt->virt_hdr, 0, sizeof(pkt->virt_hdr));
+}
+
bool net_rx_pkt_is_vlan_stripped(struct NetRxPkt *pkt)
{
assert(pkt);
diff --git a/hw/net/net_rx_pkt.h b/hw/net/net_rx_pkt.h
index 7277907a22..8b69ddb2da 100644
--- a/hw/net/net_rx_pkt.h
+++ b/hw/net/net_rx_pkt.h
@@ -312,6 +312,14 @@ void net_rx_pkt_set_vhdr(struct NetRxPkt *pkt,
void net_rx_pkt_set_vhdr_iovec(struct NetRxPkt *pkt,
const struct iovec *iov, int iovcnt);
+/**
+ * unset vhdr data from packet context
+ *
+ * @pkt: packet
+ *
+ */
+void net_rx_pkt_unset_vhdr(struct NetRxPkt *pkt);
+
/**
* save packet type in packet context
*
diff --git a/hw/net/net_tx_pkt.c b/hw/net/net_tx_pkt.c
index cf46c8457f..6afd3f6693 100644
--- a/hw/net/net_tx_pkt.c
+++ b/hw/net/net_tx_pkt.c
@@ -53,8 +53,6 @@ struct NetTxPkt {
uint16_t hdr_len;
eth_pkt_types_e packet_type;
uint8_t l4proto;
-
- bool is_loopback;
};
void net_tx_pkt_init(struct NetTxPkt **pkt, PCIDevice *pci_dev,
@@ -508,12 +506,6 @@ static void net_tx_pkt_do_sw_csum(struct NetTxPkt *pkt)
iov_from_buf(iov, iov_len, csum_offset, &csum, sizeof csum);
}
-enum {
- NET_TX_PKT_FRAGMENT_L2_HDR_POS = 0,
- NET_TX_PKT_FRAGMENT_L3_HDR_POS,
- NET_TX_PKT_FRAGMENT_HEADER_NUM
-};
-
#define NET_MAX_FRAG_SG_LIST (64)
static size_t net_tx_pkt_fetch_fragment(struct NetTxPkt *pkt,
@@ -522,7 +514,7 @@ static size_t net_tx_pkt_fetch_fragment(struct NetTxPkt *pkt,
size_t fetched = 0;
struct iovec *src = pkt->vec;
- *dst_idx = NET_TX_PKT_FRAGMENT_HEADER_NUM;
+ *dst_idx = NET_TX_PKT_PL_START_FRAG;
while (fetched < IP_FRAG_ALIGN_SIZE(pkt->virt_hdr.gso_size)) {
@@ -555,18 +547,22 @@ static size_t net_tx_pkt_fetch_fragment(struct NetTxPkt *pkt,
return fetched;
}
-static inline void net_tx_pkt_sendv(struct NetTxPkt *pkt,
- NetClientState *nc, const struct iovec *iov, int iov_cnt)
+static void net_tx_pkt_sendv(
+ void *opaque, const struct iovec *iov, int iov_cnt,
+ const struct iovec *virt_iov, int virt_iov_cnt)
{
- if (pkt->is_loopback) {
- qemu_receive_packet_iov(nc, iov, iov_cnt);
+ NetClientState *nc = opaque;
+
+ if (qemu_get_using_vnet_hdr(nc->peer)) {
+ qemu_sendv_packet(nc, virt_iov, virt_iov_cnt);
} else {
qemu_sendv_packet(nc, iov, iov_cnt);
}
}
static bool net_tx_pkt_do_sw_fragmentation(struct NetTxPkt *pkt,
- NetClientState *nc)
+ NetTxPktCallback callback,
+ void *context)
{
struct iovec fragment[NET_MAX_FRAG_SG_LIST];
size_t fragment_len = 0;
@@ -578,6 +574,10 @@ static bool net_tx_pkt_do_sw_fragmentation(struct NetTxPkt *pkt,
int src_idx = NET_TX_PKT_PL_START_FRAG, dst_idx;
size_t src_offset = 0;
size_t fragment_offset = 0;
+ struct virtio_net_hdr virt_hdr = {
+ .flags = pkt->virt_hdr.flags & VIRTIO_NET_HDR_F_NEEDS_CSUM ?
+ VIRTIO_NET_HDR_F_DATA_VALID : 0
+ };
l2_iov_base = pkt->vec[NET_TX_PKT_L2HDR_FRAG].iov_base;
l2_iov_len = pkt->vec[NET_TX_PKT_L2HDR_FRAG].iov_len;
@@ -585,10 +585,12 @@ static bool net_tx_pkt_do_sw_fragmentation(struct NetTxPkt *pkt,
l3_iov_len = pkt->vec[NET_TX_PKT_L3HDR_FRAG].iov_len;
/* Copy headers */
- fragment[NET_TX_PKT_FRAGMENT_L2_HDR_POS].iov_base = l2_iov_base;
- fragment[NET_TX_PKT_FRAGMENT_L2_HDR_POS].iov_len = l2_iov_len;
- fragment[NET_TX_PKT_FRAGMENT_L3_HDR_POS].iov_base = l3_iov_base;
- fragment[NET_TX_PKT_FRAGMENT_L3_HDR_POS].iov_len = l3_iov_len;
+ fragment[NET_TX_PKT_VHDR_FRAG].iov_base = &virt_hdr;
+ fragment[NET_TX_PKT_VHDR_FRAG].iov_len = sizeof(virt_hdr);
+ fragment[NET_TX_PKT_L2HDR_FRAG].iov_base = l2_iov_base;
+ fragment[NET_TX_PKT_L2HDR_FRAG].iov_len = l2_iov_len;
+ fragment[NET_TX_PKT_L3HDR_FRAG].iov_base = l3_iov_base;
+ fragment[NET_TX_PKT_L3HDR_FRAG].iov_len = l3_iov_len;
/* Put as much data as possible and send */
@@ -603,7 +605,9 @@ static bool net_tx_pkt_do_sw_fragmentation(struct NetTxPkt *pkt,
eth_fix_ip4_checksum(l3_iov_base, l3_iov_len);
- net_tx_pkt_sendv(pkt, nc, fragment, dst_idx);
+ callback(context,
+ fragment + NET_TX_PKT_L2HDR_FRAG, dst_idx - NET_TX_PKT_L2HDR_FRAG,
+ fragment + NET_TX_PKT_VHDR_FRAG, dst_idx - NET_TX_PKT_VHDR_FRAG);
fragment_offset += fragment_len;
@@ -614,12 +618,16 @@ static bool net_tx_pkt_do_sw_fragmentation(struct NetTxPkt *pkt,
bool net_tx_pkt_send(struct NetTxPkt *pkt, NetClientState *nc)
{
- bool using_vnet_hdr = qemu_get_using_vnet_hdr(nc->peer);
+ bool offload = qemu_get_using_vnet_hdr(nc->peer);
+ return net_tx_pkt_send_custom(pkt, offload, net_tx_pkt_sendv, nc);
+}
+bool net_tx_pkt_send_custom(struct NetTxPkt *pkt, bool offload,
+ NetTxPktCallback callback, void *context)
+{
assert(pkt);
- if (!using_vnet_hdr &&
- pkt->virt_hdr.flags & VIRTIO_NET_HDR_F_NEEDS_CSUM) {
+ if (!offload && pkt->virt_hdr.flags & VIRTIO_NET_HDR_F_NEEDS_CSUM) {
net_tx_pkt_do_sw_csum(pkt);
}
@@ -635,28 +643,16 @@ bool net_tx_pkt_send(struct NetTxPkt *pkt, NetClientState *nc)
}
}
- if (using_vnet_hdr ||
- pkt->virt_hdr.gso_type == VIRTIO_NET_HDR_GSO_NONE) {
- int index = using_vnet_hdr ?
- NET_TX_PKT_VHDR_FRAG : NET_TX_PKT_L2HDR_FRAG;
+ if (offload || pkt->virt_hdr.gso_type == VIRTIO_NET_HDR_GSO_NONE) {
net_tx_pkt_fix_ip6_payload_len(pkt);
- net_tx_pkt_sendv(pkt, nc, pkt->vec + index,
- pkt->payload_frags + NET_TX_PKT_PL_START_FRAG - index);
+ callback(context, pkt->vec + NET_TX_PKT_L2HDR_FRAG,
+ pkt->payload_frags + NET_TX_PKT_PL_START_FRAG - NET_TX_PKT_L2HDR_FRAG,
+ pkt->vec + NET_TX_PKT_VHDR_FRAG,
+ pkt->payload_frags + NET_TX_PKT_PL_START_FRAG - NET_TX_PKT_VHDR_FRAG);
return true;
}
- return net_tx_pkt_do_sw_fragmentation(pkt, nc);
-}
-
-bool net_tx_pkt_send_loopback(struct NetTxPkt *pkt, NetClientState *nc)
-{
- bool res;
-
- pkt->is_loopback = true;
- res = net_tx_pkt_send(pkt, nc);
- pkt->is_loopback = false;
-
- return res;
+ return net_tx_pkt_do_sw_fragmentation(pkt, callback, context);
}
void net_tx_pkt_fix_ip6_payload_len(struct NetTxPkt *pkt)
diff --git a/hw/net/net_tx_pkt.h b/hw/net/net_tx_pkt.h
index 8d3faa42fb..f57b4e034b 100644
--- a/hw/net/net_tx_pkt.h
+++ b/hw/net/net_tx_pkt.h
@@ -26,6 +26,8 @@
struct NetTxPkt;
+typedef void (* NetTxPktCallback)(void *, const struct iovec *, int, const struct iovec *, int);
+
/**
* Init function for tx packet functionality
*
@@ -161,15 +163,16 @@ void net_tx_pkt_reset(struct NetTxPkt *pkt);
bool net_tx_pkt_send(struct NetTxPkt *pkt, NetClientState *nc);
/**
-* Redirect packet directly to receive path (emulate loopback phy).
-* Handles sw offloads if vhdr is not supported.
-*
-* @pkt: packet
-* @nc: NetClientState
-* @ret: operation result
-*
-*/
-bool net_tx_pkt_send_loopback(struct NetTxPkt *pkt, NetClientState *nc);
+ * Send packet with a custom function.
+ *
+ * @pkt: packet
+ * @offload: whether the callback implements offloading
+ * @callback: a function to be called back for each transformed packet
+ * @context: a pointer to be passed to the callback.
+ * @ret: operation result
+ */
+bool net_tx_pkt_send_custom(struct NetTxPkt *pkt, bool offload,
+ NetTxPktCallback callback, void *context);
/**
* parse raw packet data and analyze offload requirements.
--
2.39.1
^ permalink raw reply related [flat|nested] 33+ messages in thread* [PATCH v4 24/28] hw/net/net_tx_pkt: Implement TCP segmentation
2023-01-30 13:46 [PATCH v4 00/28] e1000x cleanups (preliminary for IGB) Akihiko Odaki
` (22 preceding siblings ...)
2023-01-30 13:47 ` [PATCH v4 23/28] e1000e: Perform software segmentation for loopback Akihiko Odaki
@ 2023-01-30 13:47 ` Akihiko Odaki
2023-01-30 13:47 ` [PATCH v4 25/28] hw/net/net_tx_pkt: Check the payload length Akihiko Odaki
` (3 subsequent siblings)
27 siblings, 0 replies; 33+ messages in thread
From: Akihiko Odaki @ 2023-01-30 13:47 UTC (permalink / raw)
Cc: Jason Wang, Dmitry Fleytman, Thomas Huth, Laurent Vivier,
Paolo Bonzini, qemu-devel, qemu-ppc, Michael S . Tsirkin,
Yan Vugenfirer, Yuri Benditovich, Sriram Yagnaraman,
Alexander Bulekov, Akihiko Odaki
There was no proper implementation of TCP segmentation before this
change, and net_tx_pkt relied solely on IPv4 fragmentation. Not only
this is not aligned with the specification, but it also resulted in
corrupted IPv6 packets.
This is particularly problematic for the igb, a new proposed device
implementation; igb provides loopback feature for VMDq and the feature
relies on software segmentation.
Implement proper TCP segmentation in net_tx_pkt to fix such a scenario.
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
---
hw/net/net_tx_pkt.c | 248 ++++++++++++++++++++++++++++++++++++--------
include/net/eth.h | 5 -
net/eth.c | 27 -----
3 files changed, 206 insertions(+), 74 deletions(-)
diff --git a/hw/net/net_tx_pkt.c b/hw/net/net_tx_pkt.c
index 6afd3f6693..4a35e8429d 100644
--- a/hw/net/net_tx_pkt.c
+++ b/hw/net/net_tx_pkt.c
@@ -326,7 +326,8 @@ bool net_tx_pkt_build_vheader(struct NetTxPkt *pkt, bool tso_enable,
case VIRTIO_NET_HDR_GSO_TCPV6:
bytes_read = iov_to_buf(&pkt->vec[NET_TX_PKT_PL_START_FRAG],
pkt->payload_frags, 0, &l4hdr, sizeof(l4hdr));
- if (bytes_read < sizeof(l4hdr)) {
+ if (bytes_read < sizeof(l4hdr) ||
+ l4hdr.th_off * sizeof(uint32_t) < sizeof(l4hdr)) {
return false;
}
@@ -466,15 +467,14 @@ void net_tx_pkt_reset(struct NetTxPkt *pkt)
pkt->l4proto = 0;
}
-static void net_tx_pkt_do_sw_csum(struct NetTxPkt *pkt)
+static void net_tx_pkt_do_sw_csum(struct NetTxPkt *pkt,
+ struct iovec *iov, uint32_t iov_len,
+ uint16_t csl)
{
- struct iovec *iov = &pkt->vec[NET_TX_PKT_L2HDR_FRAG];
uint32_t csum_cntr;
uint16_t csum = 0;
uint32_t cso;
/* num of iovec without vhdr */
- uint32_t iov_len = pkt->payload_frags + NET_TX_PKT_PL_START_FRAG - 1;
- uint16_t csl;
size_t csum_offset = pkt->virt_hdr.csum_start + pkt->virt_hdr.csum_offset;
uint16_t l3_proto = eth_get_l3_proto(iov, 1, iov->iov_len);
@@ -482,8 +482,6 @@ static void net_tx_pkt_do_sw_csum(struct NetTxPkt *pkt)
iov_from_buf(iov, iov_len, csum_offset, &csum, sizeof csum);
/* Calculate L4 TCP/UDP checksum */
- csl = pkt->payload_len;
-
csum_cntr = 0;
cso = 0;
/* add pseudo header to csum */
@@ -509,14 +507,13 @@ static void net_tx_pkt_do_sw_csum(struct NetTxPkt *pkt)
#define NET_MAX_FRAG_SG_LIST (64)
static size_t net_tx_pkt_fetch_fragment(struct NetTxPkt *pkt,
- int *src_idx, size_t *src_offset, struct iovec *dst, int *dst_idx)
+ int *src_idx, size_t *src_offset, size_t src_len,
+ struct iovec *dst, int *dst_idx)
{
size_t fetched = 0;
struct iovec *src = pkt->vec;
- *dst_idx = NET_TX_PKT_PL_START_FRAG;
-
- while (fetched < IP_FRAG_ALIGN_SIZE(pkt->virt_hdr.gso_size)) {
+ while (fetched < src_len) {
/* no more place in fragment iov */
if (*dst_idx == NET_MAX_FRAG_SG_LIST) {
@@ -531,7 +528,7 @@ static size_t net_tx_pkt_fetch_fragment(struct NetTxPkt *pkt,
dst[*dst_idx].iov_base = src[*src_idx].iov_base + *src_offset;
dst[*dst_idx].iov_len = MIN(src[*src_idx].iov_len - *src_offset,
- IP_FRAG_ALIGN_SIZE(pkt->virt_hdr.gso_size) - fetched);
+ src_len - fetched);
*src_offset += dst[*dst_idx].iov_len;
fetched += dst[*dst_idx].iov_len;
@@ -560,58 +557,223 @@ static void net_tx_pkt_sendv(
}
}
+static bool net_tx_pkt_tcp_fragment_init(struct NetTxPkt *pkt,
+ struct iovec *fragment,
+ int *pl_idx,
+ size_t *l4hdr_len,
+ int *src_idx,
+ size_t *src_offset,
+ size_t *src_len)
+{
+ struct iovec *l4 = fragment + NET_TX_PKT_PL_START_FRAG;
+ size_t bytes_read = 0;
+ struct tcp_hdr *th;
+
+ if (!pkt->payload_frags) {
+ return false;
+ }
+
+ l4->iov_len = pkt->virt_hdr.hdr_len - pkt->hdr_len;
+ l4->iov_base = g_malloc(l4->iov_len);
+
+ *src_idx = NET_TX_PKT_PL_START_FRAG;
+ while (pkt->vec[*src_idx].iov_len < l4->iov_len - bytes_read) {
+ memcpy((char *)l4->iov_base + bytes_read, pkt->vec[*src_idx].iov_base,
+ pkt->vec[*src_idx].iov_len);
+
+ bytes_read += pkt->vec[*src_idx].iov_len;
+
+ (*src_idx)++;
+ if (*src_idx >= pkt->payload_frags + NET_TX_PKT_PL_START_FRAG) {
+ g_free(l4->iov_base);
+ return false;
+ }
+ }
+
+ *src_offset = l4->iov_len - bytes_read;
+ memcpy((char *)l4->iov_base + bytes_read, pkt->vec[*src_idx].iov_base,
+ *src_offset);
+
+ th = l4->iov_base;
+ th->th_flags &= ~(TH_FIN | TH_PUSH);
+
+ *pl_idx = NET_TX_PKT_PL_START_FRAG + 1;
+ *l4hdr_len = l4->iov_len;
+ *src_len = pkt->virt_hdr.gso_size;
+
+ return true;
+}
+
+static void net_tx_pkt_tcp_fragment_deinit(struct iovec *fragment)
+{
+ g_free(fragment[NET_TX_PKT_PL_START_FRAG].iov_base);
+}
+
+static void net_tx_pkt_tcp_fragment_fix(struct NetTxPkt *pkt,
+ struct iovec *fragment,
+ size_t fragment_len,
+ uint8_t gso_type)
+{
+ struct iovec *l3hdr = fragment + NET_TX_PKT_L3HDR_FRAG;
+ struct iovec *l4hdr = fragment + NET_TX_PKT_PL_START_FRAG;
+ struct ip_header *ip = l3hdr->iov_base;
+ struct ip6_header *ip6 = l3hdr->iov_base;
+ size_t len = l3hdr->iov_len + l4hdr->iov_len + fragment_len;
+
+ switch (gso_type) {
+ case VIRTIO_NET_HDR_GSO_TCPV4:
+ ip->ip_len = cpu_to_be16(len);
+ eth_fix_ip4_checksum(l3hdr->iov_base, l3hdr->iov_len);
+ break;
+
+ case VIRTIO_NET_HDR_GSO_TCPV6:
+ len -= sizeof(struct ip6_header);
+ ip6->ip6_ctlun.ip6_un1.ip6_un1_plen = cpu_to_be16(len);
+ break;
+ }
+}
+
+static void net_tx_pkt_tcp_fragment_advance(struct NetTxPkt *pkt,
+ struct iovec *fragment,
+ size_t fragment_len,
+ uint8_t gso_type)
+{
+ struct iovec *l3hdr = fragment + NET_TX_PKT_L3HDR_FRAG;
+ struct iovec *l4hdr = fragment + NET_TX_PKT_PL_START_FRAG;
+ struct ip_header *ip = l3hdr->iov_base;
+ struct tcp_hdr *th = l4hdr->iov_base;
+
+ if (gso_type == VIRTIO_NET_HDR_GSO_TCPV4) {
+ ip->ip_id = cpu_to_be16(be16_to_cpu(ip->ip_id) + 1);
+ }
+
+ th->th_seq = cpu_to_be32(be32_to_cpu(th->th_seq) + fragment_len);
+ th->th_flags &= ~TH_CWR;
+}
+
+static void net_tx_pkt_udp_fragment_init(struct NetTxPkt *pkt,
+ int *pl_idx,
+ size_t *l4hdr_len,
+ int *src_idx, size_t *src_offset,
+ size_t *src_len)
+{
+ *pl_idx = NET_TX_PKT_PL_START_FRAG;
+ *l4hdr_len = 0;
+ *src_idx = NET_TX_PKT_PL_START_FRAG;
+ *src_offset = 0;
+ *src_len = IP_FRAG_ALIGN_SIZE(pkt->virt_hdr.gso_size);
+}
+
+static void net_tx_pkt_udp_fragment_fix(struct NetTxPkt *pkt,
+ struct iovec *fragment,
+ size_t fragment_offset,
+ size_t fragment_len)
+{
+ bool more_frags = fragment_offset + fragment_len < pkt->payload_len;
+ uint16_t orig_flags;
+ struct iovec *l3hdr = fragment + NET_TX_PKT_L3HDR_FRAG;
+ struct ip_header *ip = l3hdr->iov_base;
+ uint16_t frag_off_units = fragment_offset / IP_FRAG_UNIT_SIZE;
+ uint16_t new_ip_off;
+
+ assert(fragment_offset % IP_FRAG_UNIT_SIZE == 0);
+ assert((frag_off_units & ~IP_OFFMASK) == 0);
+
+ orig_flags = be16_to_cpu(ip->ip_off) & ~(IP_OFFMASK | IP_MF);
+ new_ip_off = frag_off_units | orig_flags | (more_frags ? IP_MF : 0);
+ ip->ip_off = cpu_to_be16(new_ip_off);
+ ip->ip_len = cpu_to_be16(l3hdr->iov_len + fragment_len);
+
+ eth_fix_ip4_checksum(l3hdr->iov_base, l3hdr->iov_len);
+}
+
static bool net_tx_pkt_do_sw_fragmentation(struct NetTxPkt *pkt,
NetTxPktCallback callback,
void *context)
{
+ uint8_t gso_type = pkt->virt_hdr.gso_type & ~VIRTIO_NET_HDR_GSO_ECN;
+
struct iovec fragment[NET_MAX_FRAG_SG_LIST];
- size_t fragment_len = 0;
- bool more_frags = false;
-
- /* some pointers for shorter code */
- void *l2_iov_base, *l3_iov_base;
- size_t l2_iov_len, l3_iov_len;
- int src_idx = NET_TX_PKT_PL_START_FRAG, dst_idx;
- size_t src_offset = 0;
+ size_t fragment_len;
+ size_t l4hdr_len;
+ size_t src_len;
+
+ int src_idx, dst_idx, pl_idx;
+ size_t src_offset;
size_t fragment_offset = 0;
struct virtio_net_hdr virt_hdr = {
.flags = pkt->virt_hdr.flags & VIRTIO_NET_HDR_F_NEEDS_CSUM ?
VIRTIO_NET_HDR_F_DATA_VALID : 0
};
- l2_iov_base = pkt->vec[NET_TX_PKT_L2HDR_FRAG].iov_base;
- l2_iov_len = pkt->vec[NET_TX_PKT_L2HDR_FRAG].iov_len;
- l3_iov_base = pkt->vec[NET_TX_PKT_L3HDR_FRAG].iov_base;
- l3_iov_len = pkt->vec[NET_TX_PKT_L3HDR_FRAG].iov_len;
-
/* Copy headers */
fragment[NET_TX_PKT_VHDR_FRAG].iov_base = &virt_hdr;
fragment[NET_TX_PKT_VHDR_FRAG].iov_len = sizeof(virt_hdr);
- fragment[NET_TX_PKT_L2HDR_FRAG].iov_base = l2_iov_base;
- fragment[NET_TX_PKT_L2HDR_FRAG].iov_len = l2_iov_len;
- fragment[NET_TX_PKT_L3HDR_FRAG].iov_base = l3_iov_base;
- fragment[NET_TX_PKT_L3HDR_FRAG].iov_len = l3_iov_len;
+ fragment[NET_TX_PKT_L2HDR_FRAG] = pkt->vec[NET_TX_PKT_L2HDR_FRAG];
+ fragment[NET_TX_PKT_L3HDR_FRAG] = pkt->vec[NET_TX_PKT_L3HDR_FRAG];
+ switch (gso_type) {
+ case VIRTIO_NET_HDR_GSO_TCPV4:
+ case VIRTIO_NET_HDR_GSO_TCPV6:
+ if (!net_tx_pkt_tcp_fragment_init(pkt, fragment, &pl_idx, &l4hdr_len,
+ &src_idx, &src_offset, &src_len)) {
+ return false;
+ }
+ break;
- /* Put as much data as possible and send */
- do {
- fragment_len = net_tx_pkt_fetch_fragment(pkt, &src_idx, &src_offset,
- fragment, &dst_idx);
+ case VIRTIO_NET_HDR_GSO_UDP:
+ net_tx_pkt_do_sw_csum(pkt, &pkt->vec[NET_TX_PKT_L2HDR_FRAG],
+ pkt->payload_frags + NET_TX_PKT_PL_START_FRAG - 1,
+ pkt->payload_len);
+ net_tx_pkt_udp_fragment_init(pkt, &pl_idx, &l4hdr_len,
+ &src_idx, &src_offset, &src_len);
+ break;
- more_frags = (fragment_offset + fragment_len < pkt->payload_len);
+ default:
+ abort();
+ }
- eth_setup_ip4_fragmentation(l2_iov_base, l2_iov_len, l3_iov_base,
- l3_iov_len, fragment_len, fragment_offset, more_frags);
+ /* Put as much data as possible and send */
+ while (true) {
+ dst_idx = pl_idx;
+ fragment_len = net_tx_pkt_fetch_fragment(pkt,
+ &src_idx, &src_offset, src_len, fragment, &dst_idx);
+ if (!fragment_len) {
+ break;
+ }
- eth_fix_ip4_checksum(l3_iov_base, l3_iov_len);
+ switch (gso_type) {
+ case VIRTIO_NET_HDR_GSO_TCPV4:
+ case VIRTIO_NET_HDR_GSO_TCPV6:
+ net_tx_pkt_tcp_fragment_fix(pkt, fragment, fragment_len, gso_type);
+ net_tx_pkt_do_sw_csum(pkt, fragment + NET_TX_PKT_L2HDR_FRAG,
+ dst_idx - NET_TX_PKT_L2HDR_FRAG,
+ l4hdr_len + fragment_len);
+ break;
+
+ case VIRTIO_NET_HDR_GSO_UDP:
+ net_tx_pkt_udp_fragment_fix(pkt, fragment, fragment_offset,
+ fragment_len);
+ break;
+ }
callback(context,
fragment + NET_TX_PKT_L2HDR_FRAG, dst_idx - NET_TX_PKT_L2HDR_FRAG,
fragment + NET_TX_PKT_VHDR_FRAG, dst_idx - NET_TX_PKT_VHDR_FRAG);
+ if (gso_type == VIRTIO_NET_HDR_GSO_TCPV4 ||
+ gso_type == VIRTIO_NET_HDR_GSO_TCPV6) {
+ net_tx_pkt_tcp_fragment_advance(pkt, fragment, fragment_len,
+ gso_type);
+ }
+
fragment_offset += fragment_len;
+ }
- } while (fragment_len && more_frags);
+ if (gso_type == VIRTIO_NET_HDR_GSO_TCPV4 ||
+ gso_type == VIRTIO_NET_HDR_GSO_TCPV6) {
+ net_tx_pkt_tcp_fragment_deinit(fragment);
+ }
return true;
}
@@ -627,10 +789,6 @@ bool net_tx_pkt_send_custom(struct NetTxPkt *pkt, bool offload,
{
assert(pkt);
- if (!offload && pkt->virt_hdr.flags & VIRTIO_NET_HDR_F_NEEDS_CSUM) {
- net_tx_pkt_do_sw_csum(pkt);
- }
-
/*
* Since underlying infrastructure does not support IP datagrams longer
* than 64K we should drop such packets and don't even try to send
@@ -644,6 +802,12 @@ bool net_tx_pkt_send_custom(struct NetTxPkt *pkt, bool offload,
}
if (offload || pkt->virt_hdr.gso_type == VIRTIO_NET_HDR_GSO_NONE) {
+ if (!offload && pkt->virt_hdr.flags & VIRTIO_NET_HDR_F_NEEDS_CSUM) {
+ net_tx_pkt_do_sw_csum(pkt, &pkt->vec[NET_TX_PKT_L2HDR_FRAG],
+ pkt->payload_frags + NET_TX_PKT_PL_START_FRAG - 1,
+ pkt->payload_len);
+ }
+
net_tx_pkt_fix_ip6_payload_len(pkt);
callback(context, pkt->vec + NET_TX_PKT_L2HDR_FRAG,
pkt->payload_frags + NET_TX_PKT_PL_START_FRAG - NET_TX_PKT_L2HDR_FRAG,
diff --git a/include/net/eth.h b/include/net/eth.h
index 6e699b0d7a..2b4374fae4 100644
--- a/include/net/eth.h
+++ b/include/net/eth.h
@@ -400,11 +400,6 @@ void eth_get_protocols(const struct iovec *iov, int iovcnt,
eth_ip4_hdr_info *ip4hdr_info,
eth_l4_hdr_info *l4hdr_info);
-void eth_setup_ip4_fragmentation(const void *l2hdr, size_t l2hdr_len,
- void *l3hdr, size_t l3hdr_len,
- size_t l3payload_len,
- size_t frag_offset, bool more_frags);
-
void
eth_fix_ip4_checksum(void *l3hdr, size_t l3hdr_len);
diff --git a/net/eth.c b/net/eth.c
index f074b2f9f3..36d39b4357 100644
--- a/net/eth.c
+++ b/net/eth.c
@@ -314,33 +314,6 @@ eth_strip_vlan_ex(const struct iovec *iov, int iovcnt, size_t iovoff,
return 0;
}
-void
-eth_setup_ip4_fragmentation(const void *l2hdr, size_t l2hdr_len,
- void *l3hdr, size_t l3hdr_len,
- size_t l3payload_len,
- size_t frag_offset, bool more_frags)
-{
- const struct iovec l2vec = {
- .iov_base = (void *) l2hdr,
- .iov_len = l2hdr_len
- };
-
- if (eth_get_l3_proto(&l2vec, 1, l2hdr_len) == ETH_P_IP) {
- uint16_t orig_flags;
- struct ip_header *iphdr = (struct ip_header *) l3hdr;
- uint16_t frag_off_units = frag_offset / IP_FRAG_UNIT_SIZE;
- uint16_t new_ip_off;
-
- assert(frag_offset % IP_FRAG_UNIT_SIZE == 0);
- assert((frag_off_units & ~IP_OFFMASK) == 0);
-
- orig_flags = be16_to_cpu(iphdr->ip_off) & ~(IP_OFFMASK|IP_MF);
- new_ip_off = frag_off_units | orig_flags | (more_frags ? IP_MF : 0);
- iphdr->ip_off = cpu_to_be16(new_ip_off);
- iphdr->ip_len = cpu_to_be16(l3payload_len + l3hdr_len);
- }
-}
-
void
eth_fix_ip4_checksum(void *l3hdr, size_t l3hdr_len)
{
--
2.39.1
^ permalink raw reply related [flat|nested] 33+ messages in thread* [PATCH v4 25/28] hw/net/net_tx_pkt: Check the payload length
2023-01-30 13:46 [PATCH v4 00/28] e1000x cleanups (preliminary for IGB) Akihiko Odaki
` (23 preceding siblings ...)
2023-01-30 13:47 ` [PATCH v4 24/28] hw/net/net_tx_pkt: Implement TCP segmentation Akihiko Odaki
@ 2023-01-30 13:47 ` Akihiko Odaki
2023-01-30 13:47 ` [PATCH v4 26/28] e1000e: Do not assert when MSI-X is disabled later Akihiko Odaki
` (2 subsequent siblings)
27 siblings, 0 replies; 33+ messages in thread
From: Akihiko Odaki @ 2023-01-30 13:47 UTC (permalink / raw)
Cc: Jason Wang, Dmitry Fleytman, Thomas Huth, Laurent Vivier,
Paolo Bonzini, qemu-devel, qemu-ppc, Michael S . Tsirkin,
Yan Vugenfirer, Yuri Benditovich, Sriram Yagnaraman,
Alexander Bulekov, Akihiko Odaki
Check the payload length if checksumming to ensure the payload contains
the space for the resulting value.
This bug was found by Alexander Bulekov with the fuzzer:
https://patchew.org/QEMU/20230129053316.1071513-1-alxndr@bu.edu/
The fixed test case is:
fuzz/crash_6aeaa33e7211ecd603726c53e834df4c6d1e08bc
Fixes: e263cd49c7 ("Packet abstraction for VMWARE network devices")
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
---
hw/net/net_tx_pkt.c | 6 ++++++
1 file changed, 6 insertions(+)
diff --git a/hw/net/net_tx_pkt.c b/hw/net/net_tx_pkt.c
index 4a35e8429d..986a3adfe9 100644
--- a/hw/net/net_tx_pkt.c
+++ b/hw/net/net_tx_pkt.c
@@ -342,11 +342,17 @@ bool net_tx_pkt_build_vheader(struct NetTxPkt *pkt, bool tso_enable,
if (csum_enable) {
switch (pkt->l4proto) {
case IP_PROTO_TCP:
+ if (pkt->payload_len < sizeof(struct tcp_hdr)) {
+ return false;
+ }
pkt->virt_hdr.flags = VIRTIO_NET_HDR_F_NEEDS_CSUM;
pkt->virt_hdr.csum_start = pkt->hdr_len;
pkt->virt_hdr.csum_offset = offsetof(struct tcp_hdr, th_sum);
break;
case IP_PROTO_UDP:
+ if (pkt->payload_len < sizeof(struct udp_hdr)) {
+ return false;
+ }
pkt->virt_hdr.flags = VIRTIO_NET_HDR_F_NEEDS_CSUM;
pkt->virt_hdr.csum_start = pkt->hdr_len;
pkt->virt_hdr.csum_offset = offsetof(struct udp_hdr, uh_sum);
--
2.39.1
^ permalink raw reply related [flat|nested] 33+ messages in thread* [PATCH v4 26/28] e1000e: Do not assert when MSI-X is disabled later
2023-01-30 13:46 [PATCH v4 00/28] e1000x cleanups (preliminary for IGB) Akihiko Odaki
` (24 preceding siblings ...)
2023-01-30 13:47 ` [PATCH v4 25/28] hw/net/net_tx_pkt: Check the payload length Akihiko Odaki
@ 2023-01-30 13:47 ` Akihiko Odaki
2023-01-30 13:47 ` [PATCH v4 27/28] MAINTAINERS: Add Akihiko Odaki as a e1000e reviewer Akihiko Odaki
2023-01-30 13:47 ` [PATCH v4 28/28] MAINTAINERS: Add e1000e test files Akihiko Odaki
27 siblings, 0 replies; 33+ messages in thread
From: Akihiko Odaki @ 2023-01-30 13:47 UTC (permalink / raw)
Cc: Jason Wang, Dmitry Fleytman, Thomas Huth, Laurent Vivier,
Paolo Bonzini, qemu-devel, qemu-ppc, Michael S . Tsirkin,
Yan Vugenfirer, Yuri Benditovich, Sriram Yagnaraman,
Alexander Bulekov, Akihiko Odaki
Assertions will fail if MSI-X gets disabled while a timer for MSI-X
interrupts is running so remove them to avoid abortions. Fortunately,
nothing bad happens even if the assertions won't trigger as
msix_notify(), called by timer handlers, does nothing when MSI-X is
disabled.
This bug was found by Alexander Bulekov when fuzzing igb, a new
device implementation derived from e1000e:
https://patchew.org/QEMU/20230129053316.1071513-1-alxndr@bu.edu/
The fixed test case is:
fuzz/crash_aea040166819193cf9fedb810c6d100221da721a
Fixes: 6f3fbe4ed0 ("net: Introduce e1000e device emulation")
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
---
hw/net/e1000e_core.c | 4 ----
1 file changed, 4 deletions(-)
diff --git a/hw/net/e1000e_core.c b/hw/net/e1000e_core.c
index ff93547f88..76c7814cb8 100644
--- a/hw/net/e1000e_core.c
+++ b/hw/net/e1000e_core.c
@@ -162,8 +162,6 @@ e1000e_intrmgr_on_throttling_timer(void *opaque)
{
E1000IntrDelayTimer *timer = opaque;
- assert(!msix_enabled(timer->core->owner));
-
timer->running = false;
if (msi_enabled(timer->core->owner)) {
@@ -183,8 +181,6 @@ e1000e_intrmgr_on_msix_throttling_timer(void *opaque)
E1000IntrDelayTimer *timer = opaque;
int idx = timer - &timer->core->eitr[0];
- assert(msix_enabled(timer->core->owner));
-
timer->running = false;
trace_e1000e_irq_msix_notify_postponed_vec(idx);
--
2.39.1
^ permalink raw reply related [flat|nested] 33+ messages in thread* [PATCH v4 27/28] MAINTAINERS: Add Akihiko Odaki as a e1000e reviewer
2023-01-30 13:46 [PATCH v4 00/28] e1000x cleanups (preliminary for IGB) Akihiko Odaki
` (25 preceding siblings ...)
2023-01-30 13:47 ` [PATCH v4 26/28] e1000e: Do not assert when MSI-X is disabled later Akihiko Odaki
@ 2023-01-30 13:47 ` Akihiko Odaki
2023-01-30 13:47 ` [PATCH v4 28/28] MAINTAINERS: Add e1000e test files Akihiko Odaki
27 siblings, 0 replies; 33+ messages in thread
From: Akihiko Odaki @ 2023-01-30 13:47 UTC (permalink / raw)
Cc: Jason Wang, Dmitry Fleytman, Thomas Huth, Laurent Vivier,
Paolo Bonzini, qemu-devel, qemu-ppc, Michael S . Tsirkin,
Yan Vugenfirer, Yuri Benditovich, Sriram Yagnaraman,
Alexander Bulekov, Akihiko Odaki
I want to know to be notified when there is a new change for e1000e
as e1000e is similar to igb and such a change may also be applicable for
igb.
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
---
MAINTAINERS | 2 ++
1 file changed, 2 insertions(+)
diff --git a/MAINTAINERS b/MAINTAINERS
index 08ad1e5341..958915f227 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -2208,11 +2208,13 @@ F: docs/specs/rocker.txt
e1000x
M: Dmitry Fleytman <dmitry.fleytman@gmail.com>
+R: Akihiko Odaki <akihiko.odaki@daynix.com>
S: Maintained
F: hw/net/e1000x*
e1000e
M: Dmitry Fleytman <dmitry.fleytman@gmail.com>
+R: Akihiko Odaki <akihiko.odaki@daynix.com>
S: Maintained
F: hw/net/e1000e*
F: tests/qtest/fuzz-e1000e-test.c
--
2.39.1
^ permalink raw reply related [flat|nested] 33+ messages in thread* [PATCH v4 28/28] MAINTAINERS: Add e1000e test files
2023-01-30 13:46 [PATCH v4 00/28] e1000x cleanups (preliminary for IGB) Akihiko Odaki
` (26 preceding siblings ...)
2023-01-30 13:47 ` [PATCH v4 27/28] MAINTAINERS: Add Akihiko Odaki as a e1000e reviewer Akihiko Odaki
@ 2023-01-30 13:47 ` Akihiko Odaki
27 siblings, 0 replies; 33+ messages in thread
From: Akihiko Odaki @ 2023-01-30 13:47 UTC (permalink / raw)
Cc: Jason Wang, Dmitry Fleytman, Thomas Huth, Laurent Vivier,
Paolo Bonzini, qemu-devel, qemu-ppc, Michael S . Tsirkin,
Yan Vugenfirer, Yuri Benditovich, Sriram Yagnaraman,
Alexander Bulekov, Akihiko Odaki
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Acked-by: Thomas Huth <thuth@redhat.com>
---
MAINTAINERS | 2 ++
1 file changed, 2 insertions(+)
diff --git a/MAINTAINERS b/MAINTAINERS
index 958915f227..e920d0061e 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -2218,6 +2218,8 @@ R: Akihiko Odaki <akihiko.odaki@daynix.com>
S: Maintained
F: hw/net/e1000e*
F: tests/qtest/fuzz-e1000e-test.c
+F: tests/qtest/e1000e-test.c
+F: tests/qtest/libqos/e1000e.*
eepro100
M: Stefan Weil <sw@weilnetz.de>
--
2.39.1
^ permalink raw reply related [flat|nested] 33+ messages in thread