2019-12-03
This release note applies to driver package nt_driver_3gd-linux-3.17.2
ID | Customer tracking number | Summary | Implemented in |
---|---|---|---|
43198 | Support for Capture Flow Management FPGAs | 3.17.2 |
ID | Customer tracking number | Summary | Found in version | Resolved in version |
---|---|---|---|---|
43154 | NIM sensor reading for SFP's without DMI data e.g. trispeed NIM's will show constant values from reading non-existing sensor values. | 3.15.0 | 3.17.1 | |
43151 | 43143 (Support) | It might happen when disabling/enabling a port that the whole system becomes unresponsive. | 3.14.0 | 3.17.1 |
43123 | 43058 (Support) | When plugging some types of SFP NIM modules e.g. AFCT-5715PZ one might experience that link is not obtained by the link partner because the laser is not turned on. Re-plugging might solve the problem but not always. | 3.14.0 | 3.17.1 |
43082 | 42873 (Support) | When having defined at least one "inline" transmission (tx) host buffer (i.e., having defined a stream), and when attempting to execute the NTPL statement "delete = all" while a program is retrieving data on the stream, the service may report "Tx TBH range error on Tx feed |
3.12.1 | 3.16.1 |
43067 | 43062 (Support) | ntservice crash occuring when the a KeyMatch test uses a KeySetId reserved for a Key test. | 3.15.1 | 3.17.2 |
43065 | 43059 (Support) | The Streams view of the profiling tool displays a list of open streams/connections. If the list is empty, some key presses, e.g., pressing 'right arrow' or 'page down', will cause the tool to malfunction and stop its execution. | 3.14.0 | 3.16.1 |
43051 | 42421 (Support) | For servers with a large amount of memory and a high utilization of the system's memory and buffer caches, a Linux system may report 'cpu soft lockup' when starting Napatech driver, and in some cases the server may restart. To prevent this issue, flush the caches before starting the driver. | 3.4.1 | 3.16.1 |
ID | Customer tracking number | Summary | Found in version |
---|---|---|---|
42948 | Frequent CRC errors when using DAC cables shorter than 1 meter. | 3.13.3 | |
42936 | The link diode extinguishes when the port is disabled as it should, but lights up again when the module is unplugged and reinserted with the port still disabled. No link is up. But the diode lights up as if there is. Status is read correct from software. | 3.13.3 | |
42782 | 42636 (Support) |
NTPL expressions containing OR'ed length tests AND'ed with another expression, e.g.
Assign[StreamId = 0] =((Port == 0) AND (Length == 100 or Length == 200)) may cause a crash of NT service. Workaround: The recommended workaround is to split the NTPL statement into two statements, e.g.: Assign[StreamId = 0] =((Port == 0) AND (Length == 100)) Assign[StreamId = 0] =((Port == 0) AND (Length == 200)) |
3.10.2 |
42719 | 42701 (Support) | ntservice hanging if specifying more than 128 tx host buffers in ntservice.ini | 3.9.1 |
42653 | 42616 (fix) |
When using host loopback on NT200A01 without any NIMs plugged in, it is not possible to get link on a QSFP28-LR4 NIM inserted afterwards unless FEC has been disabled (for SR modules) in the ini file
Workaround: If QSFP28-LR module is to be used then either plug it in before start-up of the system or set DisableFEC=TRUE in ntsevice.ini |
3.11.1 |
42652 | 42616 (fix) |
FEC cannot be changed dynamically on NT200A01, if traffic has been applied. Hence, it is not possible to dynamically swap between SR and LR QSFP28 if unless FEC has been disabled for SR modules in the ini file
Workaround: If QSFP28-LR module is to be used then either plug it in before start-up of the system or set DisableFEC=TRUE in ntsevice.ini |
3.11.1 |
42257 |
Dynamic hostbuffers are not supported by Intel PAC A10 accelerators.
Therefore NTAPI Inline features are not supported by Intel PAC A10 accelerators. Workaround: Applications using DPDK will support Inline for Intel PAC A10 accelerators. |
3.9.1 | |
42256 |
The performance of the Intel PAC A10 accelerator when using 1 hostbuffer and small packets (64-65 bytes length) is limited to approximately 95% of the theoretical performance.
In all other scenarios full performance (100%) can be expected. |
3.9.1 | |
41150 | The Automatic over-temperature shutdown will only stop NT service, it cannot shutdown the Intel Programmable Acceleration Card Blue bitstream. | 3.9.0 | |
38699 |
Using NT_NetRxGet* or NT_NetRxGetMult* first time on newly created or reconfigured streams can lead to initial package drops when the system is configured with large hostbuffers or many client applications.
This will only be triggered on newly created and reconfigured streams. Workaround: If packet loss is observed, to get clean statistics no traffic should be received on connected ports while streams are being started. |
3.7.5 | |
32765 | When using the GlobalSync feature, link down/up events causes the port to enter 'unknown' state | 2.9.4 | |
14086 |
If NUMA node zero is offline and /opt/napatech3/config/ntservice.ini does not exist, the driver cannot start because it fails to allocate memory for the default host buffer set-up.
Workaround: Use an existing ntservice.ini or specify the host buffer set-up on an online NUMA node: '/opt/napatech3/bin/start.sh -o adapter0.BusId=0000:04:00.0 -o adapter0.AdapterType=NT40E3_4_PTP -o adapter0.HostBuffersRx=[4,32,1] -o adapter0.HostBuffersTx=[2,16,1]'. BusId and AdapterType are only required if there are multiple adapters in the server; the BusId can be obtained with 'lspci | grep Napa'. |
2.8.5 | |
10740 | When using a Cisco DAC pluggable, link is detected even if the other end of the cable is unplugged. | 2.7.0 | |
9189 | 9137 (Support) |
The Linux kernel v3.8->v3.12 has a bug in the NUMA balancing code which was introduced in v3.8. See https://bugzilla.kernel.org/show_bug.cgi?id=60734
The issue causes high execution delays on cores running on other than NUMA 0 if the feature is enabled. To work around the problem, disable the NUMA balancing by adding "numa_balancing=disable" to the kernel command line. The Linux kernel 3.13 received significant NUMA updates which fixes this issue. The 3.10 kernel in RHEL 7, CentOS 7 and Oracle Linux 7 (not the 3.8.13 uek kernel) contains a fix for this, thus the work around is not needed. |
0.1.1 |
8719 | Setting TimeSyncTimeJumpThreshold for OS timesync to values > 0, will enable the time jump feature and results in a jump threshold of 1 second. | 2.4.1 | |
4324 | 3GD requires at least gcc 4.0 to build and at least glibc 2.5 to run. | 1.1.0 |
Documentation is not included in this release
The Napatech driver version applies to these operating systems:
Hardware | 4GA FPGA images |
---|---|
NT20E3-2-PTP | 200-9501-02-16, 200-9501-04-04, 200-9501-06-06, 200-9501-08-06, 200-9501-09-08, 200-9501-10-07, 200-9501-15-02, 200-9501-17-02, 200-9501-18-07, 200-9501-22-05 |
NT40E3-4-PTP | 200-9502-02-16, 200-9502-04-04, 200-9502-06-06, 200-9502-06-07, 200-9502-08-06, 200-9502-08-07, 200-9502-08-08, 200-9502-09-08, 200-9502-10-07, 200-8011-22-02 |
NT80E3-2-PTP | 200-9503-02-16, 200-9503-04-04, 200-9503-06-05, 200-9503-08-06, 200-9503-08-07, 200-9503-08-08, 200-9503-09-08, 200-9503-10-07, 200-9503-10-09, 200-9503-15-02, 200-9503-17-02, 200-9503-18-07, 200-9503-20-04, 200-9503-22-05, 200-9519-10-05, 200-9519-10-07, 200-9519-15-02, 200-9519-17-02, 200-9519-18-07, 200-9519-22-05, 200-8005-10-01, 200-8005-15-02, 200-8005-17-02, 200-8005-22-00 |
NT200A01 | 200-9508-05-08, 200-9508-05-17, 200-9508-06-06, 200-9508-07-06, 200-9508-07-07, 200-9515-10-07, 200-9515-15-02, 200-9515-18-09, 200-9515-20-04, 200-9516-10-07, 200-9516-15-05, 200-9516-18-09, 200-9516-20-04, 200-9523-27-02, 200-9531-20-12, 200-8002-09-01, 200-8002-09-02, 200-8002-09-04, 200-8002-10-02, 200-8002-15-02, 200-8002-20-01, 200-9512-07-02, 200-9512-08-08, 200-9512-08-09, 200-9512-09-08, 200-9512-10-07, 200-9512-15-02, 200-9512-18-07, 200-9512-20-04, 200-9512-22-05, 200-9522-15-03, 200-9522-18-07, 200-9522-20-04, 200-9522-22-05, 200-8001-08-00, 200-8001-08-01, 200-8003-09-01, 200-8003-09-03, 200-8003-10-02, 200-8003-15-02, 200-8003-22-00, 200-8004-10-03, 200-8004-15-02, 200-8004-22-00, 200-8006-15-03, 200-8006-22-00, 200-8010-20-02 |
NT40A01-4x1 | 200-9500-06-06, 200-9500-06-07, 200-9500-08-06, 200-9500-09-08, 200-9500-10-07, 200-9500-15-02, 200-9500-18-07, 200-9500-22-05 |
NT100E3-1-PTP | 200-9504-01-12, 200-9505-02-16, 200-9505-04-04, 200-9505-06-05, 200-9505-08-06, 200-9505-08-07, 200-9505-09-08, 200-9505-10-08, 200-9505-10-09, 200-9505-15-02, 200-9505-18-07, 200-9505-20-05 |
NT40A01 | 200-9517-09-08 |
INTEL-A10 | 200-7000-12-00, 200-7000-12-02, 200-7000-12-06, 200-7000-22-05, 200-7001-12-00, 200-7001-12-03, 200-7001-12-06, 200-7001-22-05 |
NT200A02 | 200-9521-18-11, 200-9521-20-04, 200-9521-22-05, 200-9526-18-10, 200-9526-20-04, 200-9526-22-05, 200-9532-27-00, 200-9533-20-04, 200-9533-22-05, 200-9534-21-26, 200-9535-21-26, 200-8007-22-00, 200-8008-22-00, 200-8009-22-00 |
Hardware | 3GA FPGA images |
---|---|
NT4E | 200-9015-42-08, 200-9015-42-13, 200-9015-42-14 |
NT4E-STD | 200-9017-42-09, 200-9017-42-10 |
NT4E-INL | 200-9115-42-13 |
NT4E2-4-PTP | 200-9226-46-12, 200-9226-46-13, 200-9226-48-05, 200-9226-50-03, 200-9226-50-04, 200-9226-51-03, 200-9226-51-04 |
NT20E2 | 200-9220-44-10, 200-9220-44-12, 200-9220-45-06, 200-9220-46-09, 200-9220-47-03, 200-9220-50-03, 200-9220-50-04, 200-9220-50-05 |
NT20E2-PTP | 200-9227-50-03, 200-9227-51-03, 200-9227-51-04 |
NT20E3-2-PTP | 200-9233-52-13, 200-9233-53-01 |
NT40E3-4-PTP | 200-9232-50-04, 200-9232-51-04, 200-9232-51-07, 200-9232-52-13, 200-9232-53-01 |
NT40E2-1 | 200-9222-52-05 |
NT40E2-4 | 200-9221-44-13, 200-9221-50-04 |
Complete test plan.
See the Documentation Portal, WebHelp or DN-0449 for reference documentation on Napatech Software Suite.
2019-12-03