Re: CVS commit: src/sys/dev/pci/ixgbe
On 2023/10/12 14:50, SAITOH Masanobu wrote: > Module Name: src > Committed By: msaitoh > Date: Thu Oct 12 05:50:56 UTC 2023 > > Modified Files: > src/sys/dev/pci/ixgbe: ixgbe.c > > Log Message: > ixg(4): Don't print wrong error message about ixgbe_num_queues. > > Don't override the ixgbe_num_queues global variable. It's the default > value of the number of queues and should not override it because it > will be referenced by later device attach. For example, the number of > MSI-X vector is 64 on X540 and 18 on 82599. When both cards are inserted > to a machine that the number of CPU is 24 and X540 is probed earlier, > ixgbe_num_queues is overridden to 24 and the following error message is > printed when attaching 82599: > > ixg2: autoconfiguration error: ixgbe_num_queues (24) is too large, > using reduced amount (17). > > Note that the number of queues is in sc->num_queuss and referenced > by hw.ixgN.num_queues sysctl. The commit message was incorrect. - s/82599/82598/ - Worse thing can happen if a smaller number of MSI-X vector's device is attached earlier. The small number is set as the default value and the number of queues of the next device is unintentionally limited to it. > To generate a diff of this commit: > cvs rdiff -u -r1.341 -r1.342 src/sys/dev/pci/ixgbe/ixgbe.c > > Please note that diffs are not public domain; they are subject to the > copyright notices on the relevant files. > -- --- SAITOH Masanobu (msai...@execsw.org msai...@netbsd.org)
Re: CVS commit: src/sys/dev/pci/ixgbe
On 2020/12/11 14:01, SAITOH Masanobu wrote: Module Name:src Committed By: msaitoh Date: Fri Dec 11 05:01:19 UTC 2020 Modified Files: src/sys/dev/pci/ixgbe: ixgbe.c ixgbe_type.h Log Message: Don't use EIMC_OTHER bit because it's read only other than 82598. Documents say: 82598: All of bit 31(OTHER bit) of EIxx are reserved. In reality, at least EIMS_OTHER and EIMC_OTHER exist and the OTHER interrupt doesn't work without EIMS_OTHER. Other than 82598: + EICR's bit 31 is defined and other EIXX's bit 31 are reserved. + In reality, EIMS_OTHER is read only and EIMC_OTHER doesn't exist. If one of bit 29..16 is set, EIMS_OTHER is set to 1 (Note that bit 30(TCP timer isn't included)). Even if write bit 31 of EIMC to 1, it's ignored (EIMS_OTHER doesn't set). We introduced new spin mutex in ixgbe.c rev. 1.260, so it's OK to remove EIMC_OTHER stuff. We already set EIMS_OTHER in if_init(), so keep it for 82598. No functional change other than 82598. Another solution is to control bit 30..16 directly to mask/unmask interrupt instead of the mutex. TODO: Some MSI-X interrupt(LSC, module insertion/removal etc.)'s mask/unmask code between ixgbe_msix_admin() and ixgbe_handle_admin() may be wrong. It'll be fixed later. To generate a diff of this commit: cvs rdiff -u -r1.261 -r1.262 src/sys/dev/pci/ixgbe/ixgbe.c cvs rdiff -u -r1.45 -r1.46 src/sys/dev/pci/ixgbe/ixgbe_type.h Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files. -- --- SAITOH Masanobu (msai...@execsw.org msai...@netbsd.org)
Re: CVS commit: src/sys/dev/pci/ixgbe
On 2019/06/27 18:56, SAITOH Masanobu wrote: > Module Name: src > Committed By: msaitoh > Date: Thu Jun 27 09:56:39 UTC 2019 > > Modified Files: > src/sys/dev/pci/ixgbe: ixv.c > > Log Message: > Don't call set_vfta() if any VLAN is attached. s/is/is not/ > XXX pullup-8. > > > To generate a diff of this commit: > cvs rdiff -u -r1.115 -r1.116 src/sys/dev/pci/ixgbe/ixv.c > > Please note that diffs are not public domain; they are subject to the > copyright notices on the relevant files. > -- --- SAITOH Masanobu (msai...@execsw.org msai...@netbsd.org)
Re: CVS commit: src/sys/dev/pci/ixgbe
On 2019/03/13 19:02, SAITOH Masanobu wrote: Module Name:src Committed By: msaitoh Date: Wed Mar 13 10:02:13 UTC 2019 Modified Files: src/sys/dev/pci/ixgbe: ixgbe.c Log Message: - Fix a bug that the VLAN HW filter function is not correctly disabled s/VLAN HW filter/VLAN HW tagging/ when all vlan is detached. - Fix a bug that VLAN HW filter function is not correctly controlled on 82598. s/VLAN HW filter/VLAN HW tagging/ - Control VLAN HW filter function correctly. - Don't clear IXGBE_VLNCTRL_CFIEN bit When ETHERCAP_VLAN_HWFILTER is set. I think it's not required (and Linux doesn't do it). This change has no effect to NetBSD because ETHERCAP_VLAN_HWFILTER is not supported yet. To generate a diff of this commit: cvs rdiff -u -r1.176 -r1.177 src/sys/dev/pci/ixgbe/ixgbe.c Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files. -- --- SAITOH Masanobu (msai...@execsw.org msai...@netbsd.org)
Re: CVS commit: src/sys/dev/pci/ixgbe
On 2018/07/03 13:02, SAITOH Masanobu wrote: Module Name:src Committed By: msaitoh Date: Tue Jul 3 04:02:07 UTC 2018 Modified Files: src/sys/dev/pci/ixgbe: ixgbe.c Log Message: Remove nmbclusters check. A big problem is that the number of RX descriptor unexpectedly decreased from 2048 to 1024 if the number of port and/or the number of CPU is high. a2sdihtp4f: sysctl -a | grep ixg | grep rx_desc hw.ixg0.num_rx_desc = 2048 hw.ixg1.num_rx_desc = 2048 hw.ixg2.num_rx_desc = 2048 hw.ixg3.num_rx_desc = 1024 We don't use the mbuf cluster. The old code also had a bug that ixgbe_total_ports adds two every port and never decrement in the detach path. Found by hikaru@. The code was removed in FreeBSD when it switched to use iflib and OpenBSD removed the code 8 years ago. To generate a diff of this commit: cvs rdiff -u -r1.161 -r1.162 src/sys/dev/pci/ixgbe/ixgbe.c Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files. -- --- SAITOH Masanobu (msai...@execsw.org msai...@netbsd.org)
Re: CVS commit: src/sys/dev/pci/ixgbe
On 2018/05/07 16:56, Masanobu SAITOH wrote:
Hi.
On 2018/05/03 17:38, Frank Kardel wrote:
Hi,
I am now seeing that packet reception stops. tcpdump will only show outgoing
packets and no incoming packets.
ifconfig ixgX down
ifconfig ixgX up
gets things going again.
Code from 2018-02-24 was fine and I think even code shortly before this commit
was fine, but there was the mbuf used-after-free issue that hit me pretty hard.
So the system wasn't running for too long anyway.
Does that ring a bell ?
Frank
IMHO, this change doesn't cause such type of problem...
Could you show me the following information?
0) dmesg
This is required to know which chip is used and what type of
interrupt is used and configured (INTX, MSI or MSI-X).
1) ifconfig -v ixg0
It's required to know offload status, the MTU size and some
error count.
2) sysctl hw.ixg0 (or sysctl hw |grep ixg (for all ixg(4)s))
This output has some information of TX/RX ring.
3) vmstat -e[v]
Only for ixg(4):
vmstat -e[v] |grep ixg
ixg(4) has a lot of counters. I added error counters at
all of error paths. (If you found a error path which has
no counter, it's my fault.)
Only to check checksum counters:
options INET_CSUM_COUNTERS (for ip_input())
options TCP_CSUM_COUNTERS (for tcp_input())
options UDP_CSUM_COUNTERS (for udp_input())
options WM_EVENT_COUNTERS (only for wm(4))
options BGE_EVENT_COUNTERS (only for bge(4))
(No special options for ixg(4) because those counters are enabled by
default)
vmstat -e[v] |grep csum
4) If you use a virtual interface like vlan(4) or age(4). Please
s/age/agr/
show the configuration.
5) Whether you use options NET_MPSAFE or not.
Thanks.
Regards.
On 04/25/18 10:46, SAITOH Masanobu wrote:
Module Name: src
Committed By: msaitoh
Date: Wed Apr 25 08:46:19 UTC 2018
Modified Files:
src/sys/dev/pci/ixgbe: ix_txrx.c ixgbe.h ixgbe_netbsd.c ixgbe_netbsd.h
ixgbe_osdep.h
Log Message:
Don't free and reallocate bus_dmamem when it's not required. Currently,
the watchdog timer is completely broken and never fire (it's from FreeBSD
(pre iflib)). If the problem is fixed and watchdog fired, ixgbe_init() always
calls ixgbe_jcl_reinit() and it causes panic. The reason is that
ixgbe_local_timer1(it includes watchdog function) is softint and
xgbe_jcl_reinit() calls bus_dmamem*() functions. bus_dmamem*() can't be called
from interrupt context.
One of the way to prevent panic is use worqueue for the timer, but it's
not a small change. (I'll do it in future).
Another way is not reallocate dmamem if it's not required. If both the MTU
(rx_mbuf_sz in reality) and the number of RX descriptors are not changed, it's
not required to call bus_dmamem_{unmap,free}(). Even if we use workque, this
change save time of ixgbe_init().
I have a code to fix broken watchdog timer but it sometime causes watchdog
timeout, so I don't commit it yet.
To generate a diff of this commit:
cvs rdiff -u -r1.40 -r1.41 src/sys/dev/pci/ixgbe/ix_txrx.c
cvs rdiff -u -r1.45 -r1.46 src/sys/dev/pci/ixgbe/ixgbe.h
cvs rdiff -u -r1.6 -r1.7 src/sys/dev/pci/ixgbe/ixgbe_netbsd.c
cvs rdiff -u -r1.7 -r1.8 src/sys/dev/pci/ixgbe/ixgbe_netbsd.h
cvs rdiff -u -r1.21 -r1.22 src/sys/dev/pci/ixgbe/ixgbe_osdep.h
Please note that diffs are not public domain; they are subject to the
copyright notices on the relevant files.
--
---
SAITOH Masanobu (msai...@execsw.org
msai...@netbsd.org)
Re: CVS commit: src/sys/dev/pci/ixgbe
Hi.
On 2018/05/03 17:38, Frank Kardel wrote:
Hi,
I am now seeing that packet reception stops. tcpdump will only show outgoing
packets and no incoming packets.
ifconfig ixgX down
ifconfig ixgX up
gets things going again.
Code from 2018-02-24 was fine and I think even code shortly before this commit
was fine, but there was the mbuf used-after-free issue that hit me pretty hard.
So the system wasn't running for too long anyway.
Does that ring a bell ?
Frank
IMHO, this change doesn't cause such type of problem...
Could you show me the following information?
0) dmesg
This is required to know which chip is used and what type of
interrupt is used and configured (INTX, MSI or MSI-X).
1) ifconfig -v ixg0
It's required to know offload status, the MTU size and some
error count.
2) sysctl hw.ixg0 (or sysctl hw |grep ixg (for all ixg(4)s))
This output has some information of TX/RX ring.
3) vmstat -e[v]
Only for ixg(4):
vmstat -e[v] |grep ixg
ixg(4) has a lot of counters. I added error counters at
all of error paths. (If you found a error path which has
no counter, it's my fault.)
Only to check checksum counters:
options INET_CSUM_COUNTERS (for ip_input())
options TCP_CSUM_COUNTERS (for tcp_input())
options UDP_CSUM_COUNTERS (for udp_input())
options WM_EVENT_COUNTERS (only for wm(4))
options BGE_EVENT_COUNTERS (only for bge(4))
(No special options for ixg(4) because those counters are enabled by
default)
vmstat -e[v] |grep csum
4) If you use a virtual interface like vlan(4) or age(4). Please
show the configuration.
Regards.
On 04/25/18 10:46, SAITOH Masanobu wrote:
Module Name: src
Committed By: msaitoh
Date: Wed Apr 25 08:46:19 UTC 2018
Modified Files:
src/sys/dev/pci/ixgbe: ix_txrx.c ixgbe.h ixgbe_netbsd.c ixgbe_netbsd.h
ixgbe_osdep.h
Log Message:
Don't free and reallocate bus_dmamem when it's not required. Currently,
the watchdog timer is completely broken and never fire (it's from FreeBSD
(pre iflib)). If the problem is fixed and watchdog fired, ixgbe_init() always
calls ixgbe_jcl_reinit() and it causes panic. The reason is that
ixgbe_local_timer1(it includes watchdog function) is softint and
xgbe_jcl_reinit() calls bus_dmamem*() functions. bus_dmamem*() can't be called
from interrupt context.
One of the way to prevent panic is use worqueue for the timer, but it's
not a small change. (I'll do it in future).
Another way is not reallocate dmamem if it's not required. If both the MTU
(rx_mbuf_sz in reality) and the number of RX descriptors are not changed, it's
not required to call bus_dmamem_{unmap,free}(). Even if we use workque, this
change save time of ixgbe_init().
I have a code to fix broken watchdog timer but it sometime causes watchdog
timeout, so I don't commit it yet.
To generate a diff of this commit:
cvs rdiff -u -r1.40 -r1.41 src/sys/dev/pci/ixgbe/ix_txrx.c
cvs rdiff -u -r1.45 -r1.46 src/sys/dev/pci/ixgbe/ixgbe.h
cvs rdiff -u -r1.6 -r1.7 src/sys/dev/pci/ixgbe/ixgbe_netbsd.c
cvs rdiff -u -r1.7 -r1.8 src/sys/dev/pci/ixgbe/ixgbe_netbsd.h
cvs rdiff -u -r1.21 -r1.22 src/sys/dev/pci/ixgbe/ixgbe_osdep.h
Please note that diffs are not public domain; they are subject to the
copyright notices on the relevant files.
--
---
SAITOH Masanobu (msai...@execsw.org
msai...@netbsd.org)
Re: CVS commit: src/sys/dev/pci/ixgbe
Hi,
I am now seeing that packet reception stops. tcpdump will only show
outgoing packets and no incoming packets.
ifconfig ixgX down
ifconfig ixgX up
gets things going again.
Code from 2018-02-24 was fine and I think even code shortly before this
commit was fine, but there was the mbuf used-after-free issue that hit
me pretty hard. So the system wasn't running for too long anyway.
Does that ring a bell ?
Frank
On 04/25/18 10:46, SAITOH Masanobu wrote:
Module Name:src
Committed By: msaitoh
Date: Wed Apr 25 08:46:19 UTC 2018
Modified Files:
src/sys/dev/pci/ixgbe: ix_txrx.c ixgbe.h ixgbe_netbsd.c ixgbe_netbsd.h
ixgbe_osdep.h
Log Message:
Don't free and reallocate bus_dmamem when it's not required. Currently,
the watchdog timer is completely broken and never fire (it's from FreeBSD
(pre iflib)). If the problem is fixed and watchdog fired, ixgbe_init() always
calls ixgbe_jcl_reinit() and it causes panic. The reason is that
ixgbe_local_timer1(it includes watchdog function) is softint and
xgbe_jcl_reinit() calls bus_dmamem*() functions. bus_dmamem*() can't be called
from interrupt context.
One of the way to prevent panic is use worqueue for the timer, but it's
not a small change. (I'll do it in future).
Another way is not reallocate dmamem if it's not required. If both the MTU
(rx_mbuf_sz in reality) and the number of RX descriptors are not changed, it's
not required to call bus_dmamem_{unmap,free}(). Even if we use workque, this
change save time of ixgbe_init().
I have a code to fix broken watchdog timer but it sometime causes watchdog
timeout, so I don't commit it yet.
To generate a diff of this commit:
cvs rdiff -u -r1.40 -r1.41 src/sys/dev/pci/ixgbe/ix_txrx.c
cvs rdiff -u -r1.45 -r1.46 src/sys/dev/pci/ixgbe/ixgbe.h
cvs rdiff -u -r1.6 -r1.7 src/sys/dev/pci/ixgbe/ixgbe_netbsd.c
cvs rdiff -u -r1.7 -r1.8 src/sys/dev/pci/ixgbe/ixgbe_netbsd.h
cvs rdiff -u -r1.21 -r1.22 src/sys/dev/pci/ixgbe/ixgbe_osdep.h
Please note that diffs are not public domain; they are subject to the
copyright notices on the relevant files.
Re: CVS commit: src/sys/dev/pci/ixgbe
Hi,
I am now seeing that packet reception stops. tcpdump will only show
outgoing packets and no incoming packets.
ifconfig ixgX down
ifconfig ixgX up
gets things going again.
Code from 2018-02-24 was fine and I think even code shortly before this
commit was fine, but there was the mbuf used-after-free issue that hit
me pretty hard. So the system wasn't running for too long anyway.
Does that ring a bell ?
Frank
On 04/25/18 10:46, SAITOH Masanobu wrote:
Module Name:src
Committed By: msaitoh
Date: Wed Apr 25 08:46:19 UTC 2018
Modified Files:
src/sys/dev/pci/ixgbe: ix_txrx.c ixgbe.h ixgbe_netbsd.c ixgbe_netbsd.h
ixgbe_osdep.h
Log Message:
Don't free and reallocate bus_dmamem when it's not required. Currently,
the watchdog timer is completely broken and never fire (it's from FreeBSD
(pre iflib)). If the problem is fixed and watchdog fired, ixgbe_init() always
calls ixgbe_jcl_reinit() and it causes panic. The reason is that
ixgbe_local_timer1(it includes watchdog function) is softint and
xgbe_jcl_reinit() calls bus_dmamem*() functions. bus_dmamem*() can't be called
from interrupt context.
One of the way to prevent panic is use worqueue for the timer, but it's
not a small change. (I'll do it in future).
Another way is not reallocate dmamem if it's not required. If both the MTU
(rx_mbuf_sz in reality) and the number of RX descriptors are not changed, it's
not required to call bus_dmamem_{unmap,free}(). Even if we use workque, this
change save time of ixgbe_init().
I have a code to fix broken watchdog timer but it sometime causes watchdog
timeout, so I don't commit it yet.
To generate a diff of this commit:
cvs rdiff -u -r1.40 -r1.41 src/sys/dev/pci/ixgbe/ix_txrx.c
cvs rdiff -u -r1.45 -r1.46 src/sys/dev/pci/ixgbe/ixgbe.h
cvs rdiff -u -r1.6 -r1.7 src/sys/dev/pci/ixgbe/ixgbe_netbsd.c
cvs rdiff -u -r1.7 -r1.8 src/sys/dev/pci/ixgbe/ixgbe_netbsd.h
cvs rdiff -u -r1.21 -r1.22 src/sys/dev/pci/ixgbe/ixgbe_osdep.h
Please note that diffs are not public domain; they are subject to the
copyright notices on the relevant files.
Re: CVS commit: src/sys/dev/pci/ixgbe
On 2018/04/04 17:13, SAITOH Masanobu wrote:
Module Name:src
Committed By: msaitoh
Date: Wed Apr 4 08:13:07 UTC 2018
Modified Files:
src/sys/dev/pci/ixgbe: if_bypass.c if_fdir.c if_sriov.c ix_txrx.c
ixgbe.c ixgbe.h ixgbe_82598.c ixgbe_82598.h ixgbe_82599.c
ixgbe_82599.h ixgbe_api.c ixgbe_api.h ixgbe_common.c ixgbe_common.h
ixgbe_dcb.c ixgbe_dcb.h ixgbe_dcb_82598.c ixgbe_dcb_82598.h
ixgbe_dcb_82599.c ixgbe_dcb_82599.h ixgbe_fdir.h ixgbe_mbx.c
ixgbe_mbx.h ixgbe_osdep.c ixgbe_osdep.h ixgbe_phy.c ixgbe_phy.h
ixgbe_rss.h ixgbe_sriov.h ixgbe_type.h ixgbe_vf.c ixgbe_vf.h
ixgbe_x540.c ixgbe_x540.h ixv.c
Log Message:
Sync with the remaining part of FreeBSD r328265 except sfp_reinit stuff:
- Always schedule module intterrupt in ixgbe_config_link() when a device is
SFP+ based.
- Use not loop index but txr->me in ixv_initialize_{transmit,receive}_units().
It's required for VMDQ but NetBSD doesn't use it, so it's not a bug in
NetBSD.
- Simplify ixgbe_bp_wd_set(). No functional change.
- Whitespace.
+ Call ixgbe_set_phy_power() in ixgbe_init_locked() to
make sure we are not in power save mode.
--
---
SAITOH Masanobu (msai...@execsw.org
msai...@netbsd.org)
Re: CVS commit: src/sys/dev/pci/ixgbe
On 2017/11/23 0:15, SAITOH Masanobu wrote: Module Name:src Committed By: msaitoh Date: Wed Nov 22 15:15:09 UTC 2017 Modified Files: src/sys/dev/pci/ixgbe: if_bypass.c ixgbe.c ixgbe.h Log Message: Fix a bug that bypass adapter's sysctls aren't set. Tested with non-genuine X550-T2 bypass adapter: s/X550/X540/ - Call ixgbe_sysctl_instance() in ixgbe_bypass_init() to get sysctl's top node correctly. - ixgbe_init_device_features() refers adapter->hw.bus.func for bypass adapter. Call set_lan_id() to set adapter->hw.bus.func before calling ixgbe_init_device_features(). Without this, bypass sysctl's are added to both the first and second port. - Initalize node.sysctl_data before calling sysctl_lookup() to read correct value. To generate a diff of this commit: cvs rdiff -u -r1.1 -r1.2 src/sys/dev/pci/ixgbe/if_bypass.c cvs rdiff -u -r1.112 -r1.113 src/sys/dev/pci/ixgbe/ixgbe.c cvs rdiff -u -r1.27 -r1.28 src/sys/dev/pci/ixgbe/ixgbe.h Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files. -- --- SAITOH Masanobu (msai...@execsw.org msai...@netbsd.org)
Re: CVS commit: src/sys/dev/pci/ixgbe
On 2017/06/27 19:33, SAITOH Masanobu wrote: Module Name:src Committed By: msaitoh Date: Tue Jun 27 10:33:09 UTC 2017 Modified Files: src/sys/dev/pci/ixgbe: ixgbe.c Log Message: Fix a bug of ixg(4)'s media setting. Before: ifconfig ixg0 media 100baseTX -> advertise 100Mbps only ifconfig ixg0 media 1000baseT -> advertise 1Gbps and 1000Mbps (NG) s/1000Mbps/100Mbps/ ifconfig ixg0 media 10Gbase-T -> advertise all (NG) ifconfig ixg0 media auto -> advertise all After: ifconfig ixg0 media 100baseTX -> advertise 100Mbps only ifconfig ixg0 media 1000baseT -> advertise 1Gbps only ifconfig ixg0 media 10Gbase-T -> advertise 10Gbps only ifconfig ixg0 media auto -> advertise all To generate a diff of this commit: cvs rdiff -u -r1.93 -r1.94 src/sys/dev/pci/ixgbe/ixgbe.c Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files. -- --- SAITOH Masanobu (msai...@execsw.org msai...@netbsd.org)
