Re: [PATCH v3 00/19] dpaa: bus, net, dma and mempool improvements

[Stephen Hemminger <stephen@networkplumber.org>]

On Sun, 21 Jun 2026 20:52:09 +0530
Hemant Agrawal <hemant.agrawal@nxp.com> wrote:

> This series collects correctness fixes, cleanups and feature additions
> across the NXP DPAA bus, net, mempool and DMA drivers.
> 
> 1. Bus/fman infrastructure cleanups (patches 01, 02, 13)
>    - Refine fman symbol naming and fix unintended global scope.
>    - Scan the maximum BPID count from the device tree rather than
>      using a compile-time constant.
>    - Improve the DPAA bus log macro and fix bus-detection logic.
> 
> 2. BMI Tx statistics (patch 03)
>    - Extend fman_hw to read Tx BMI registers and expose the counters
>      through the xstats interface.
> 
> 3. Process-type guards (patch 04)
>    - Add secondary-process checks in the net, DMA and crypto drivers
>      to prevent segfaults when operations valid only in the primary
>      process are called from a secondary.
> 
> 4. FQ shutdown hardening (patches 05-11)
>    - Introduce helpers for qman channel and work-queue lookup so that
>      FQ teardown is driven by the FQ descriptor instead of ad-hoc
>      parameters.
>    - Add channel validation and CGR cleanup to the shutdown path.
>    - Clean up the Tx-confirmation FQ on device stop and remove a
>      redundant shutdown call from Rx queue setup.
> 
> 5. net/dpaa improvements (patches 12, 14, 15)
>    - Optimise FM de-configuration to avoid redundant portal drains.
>    - Streamline FMC MAC-type parsing.
>    - Report an error when deferred-start mode is requested (not
>      supported by the driver).
> 
> 6. mempool/dpaa (patches 16-17)
>    - Optimise multi-entry buffer-pool acquire/release operations.
>    - Release the BPID in the driver destructor to avoid resource leaks
>      across repeated bind/unbind cycles.
> 
> 7. dma/dpaa (patch 18)
>    - Add SG-list data validation and a workaround for erratum
>      ERR050757.
> 
> 8. net/dpaa ONIC support (patch 19)
>    - Add port-type checks for ONIC (Open Network Interface Card)
>      shared-Ethernet ports.
> 
> Changes in v3:
> - Fix out-of-bounds rte_memcpy in bman_release_fast (squashed into
>   patch 16): when num == 1 the source pointer bm_bufs[1] is one past
>   the array end; guard the copy with "if (num > 1)".
> 
> Gagandeep Singh (2):
>   bus/dpaa: enhance DPAA FQ shutdown
>   dma/dpaa: add SG data validation and ERR050757 fix
> 
> Hemant Agrawal (5):
>   net/dpaa: clean Tx confirmation FQ on device stop
>   net/dpaa: remove redundant FQ shutdown from Rx queue setup
>   net/dpaa: optimize FM deconfig
>   bus/dpaa: improve log macro and fix bus detection
>   net/dpaa: report error on using deferred start
> 
> Jun Yang (10):
>   bus/dpaa: refine fman naming and fix global scope
>   bus/dpaa: scan max BPID from DTS
>   drivers: add BMI Tx statistics
>   bus/dpaa: define helpers for qman channel and wq
>   drivers: shutdown DPAA FQ by fq descriptor
>   bus/dpaa: improve FQ shutdown with channel validation
>   drivers: add DPAA cgrid cleanup support
>   net/dpaa: optimize FMC MAC type parsing
>   drivers: optimize DPAA multi-entry buffer pool operations
>   drivers: release DPAA bpid on driver destructor
> 
> Prashant Gupta (1):
>   drivers: add process-type guards for secondary process
> 
> Vanshika Shukla (1):
>   net/dpaa: add ONIC port checks
> 
>  drivers/bus/dpaa/base/fman/fman.c         |  23 ++--
>  drivers/bus/dpaa/base/fman/fman_hw.c      | 108 ++++++++---------
>  drivers/bus/dpaa/base/qbman/bman.c        |  59 ++++------
>  drivers/bus/dpaa/base/qbman/bman_driver.c |  48 +++++---
>  drivers/bus/dpaa/base/qbman/qman.c        | 115 ++++++++++--------
>  drivers/bus/dpaa/base/qbman/qman.h        |  23 +++-
>  drivers/bus/dpaa/base/qbman/qman_driver.c |  29 ++++-
>  drivers/bus/dpaa/dpaa_bus.c               |  35 ++++--
>  drivers/bus/dpaa/dpaa_bus_base_symbols.c  |   4 +
>  drivers/bus/dpaa/include/fman.h           |  30 ++++-
>  drivers/bus/dpaa/include/fsl_bman.h       |  49 ++++++--
>  drivers/bus/dpaa/include/fsl_qman.h       |  22 +++-
>  drivers/crypto/dpaa_sec/dpaa_sec.c        |   3 -
>  drivers/dma/dpaa/dpaa_qdma.c              | 103 ++++++++++++----
>  drivers/mempool/dpaa/dpaa_mempool.c       |  75 ++++++++++--
>  drivers/mempool/dpaa/dpaa_mempool.h       |   3 +-
>  drivers/net/dpaa/dpaa_ethdev.c            | 132 ++++++++++++++++-----
>  drivers/net/dpaa/dpaa_ethdev.h            |  22 +++-
>  drivers/net/dpaa/dpaa_flow.c              | 137 +++++++++++++---------
>  drivers/net/dpaa/dpaa_flow.h              |   7 +-
>  drivers/net/dpaa/dpaa_fmc.c               |  73 +++++++-----
>  21 files changed, 748 insertions(+), 352 deletions(-)
> 

Clang build fails:
 ccache clang -Idrivers/libtmp_rte_net_dpaa.a.p -Idrivers -I../drivers -Idrivers/net/dpaa -I../drivers/net/dpaa -Ilib/ethdev -I../lib/ethdev -Ilib/eal/common -I../lib/eal/common -I. -I.. -Iconfig -I../config -Ilib/eal/include -I../lib/eal/include -Ilib/eal/linux/include -I../lib/eal/linux/include -Ilib/eal/x86/include -I../lib/eal/x86/include -I../kernel/linux -Ilib/eal -I../lib/eal -Ilib/kvargs -I../lib/kvargs -Ilib/log -I../lib/log -Ilib/metrics -I../lib/metrics -Ilib/telemetry -I../lib/telemetry -Ilib/argparse -I../lib/argparse -Ilib/net -I../lib/net -Ilib/mbuf -I../lib/mbuf -Ilib/mempool -I../lib/mempool -Ilib/ring -I../lib/ring -Ilib/meter -I../lib/meter -Idrivers/bus/pci -I../drivers/bus/pci -I../drivers/bus/pci/linux -Ilib/pci -I../lib/pci -Idrivers/bus/vdev -I../drivers/bus/vdev -Idrivers/mempool/dpaa -I../drivers/mempool/dpaa -Idrivers/bus/dpaa -I../drivers/bus/dpaa -I../drivers/bus/dpaa/base -I../drivers/bus/dpaa/include -I../drivers/bus/dpaa/base/qbman -Idrivers/common/dpaax -I../drivers/common/dpaax -I../drivers/common/dpaax/caamflib -Ilib/eventdev -I../lib/eventdev -Ilib/hash -I../lib/hash -Ilib/rcu -I../lib/rcu -Ilib/timer -I../lib/timer -Ilib/cryptodev -I../lib/cryptodev -Ilib/dmadev -I../lib/dmadev -Xclang -fcolor-diagnostics -pipe -D_FILE_OFFSET_BITS=64 -Wall -Winvalid-pch -Wextra -Werror -std=c11 -O2 -g -include rte_config.h -Wvla -Wcast-qual -Wcomma -Wdeprecated -Wformat -Wformat-nonliteral -Wformat-security -Wmissing-declarations -Wmissing-prototypes -Wnested-externs -Wold-style-definition -Wpointer-arith -Wshadow -Wsign-compare -Wstrict-prototypes -Wundef -Wwrite-strings -Wno-missing-field-initializers -D_GNU_SOURCE -fPIC -march=corei7 -mrtm -DALLOW_EXPERIMENTAL_API -DALLOW_INTERNAL_API -Wno-format-truncation -Wno-address-of-packed-member -Wno-vla -Wno-shadow -Wno-pointer-arith -DRTE_COMPONENT_CLASS=pmd_net -DRTE_COMPONENT_NAME=dpaa -DRTE_LOG_DEFAULT_LOGTYPE=pmd.net.dpaa -DRTE_ANNOTATE_LOCKS -Wthread-safety -MD -MQ drivers/libtmp_rte_net_dpaa.a.p/net_dpaa_dpaa_flow.c.o -MF drivers/libtmp_rte_net_dpaa.a.p/net_dpaa_dpaa_flow.c.o.d -o drivers/libtmp_rte_net_dpaa.a.p/net_dpaa_dpaa_flow.c.o -c ../drivers/net/dpaa/dpaa_flow.c
../drivers/net/dpaa/dpaa_flow.c:673:19: error: unused function 'get_tx_port_type' [-Werror,-Wunused-function]
  673 | static inline int get_tx_port_type(struct fman_if *fif)
      |                   ^~~~~~~~~~~~~~~~


Also did more detailed AI review:


Reviewed the v3 DPAA series (19 patches) against current main; applied
cleanly with git am and traced the affected paths. Issues by patch:

Patch 12 + 19: net/dpaa FM deconfig
Error: dpaa_eth_dev_close() can call dpaa_fm_deconfig() on a NULL
port_handle. Patch 12 added a second deconfig call (guarded only by
!(default_q || fmc_q)) plus an unconditional one at the end, and made
dpaa_fm_deconfig() idempotent with an early "if (!port_handle) return
0". Patch 19 removes that early return and only re-guards the
dpaa_fm_config() caller. Final close path:

if (!(default_q||fmc_q)) if (port_handle) deconfig  /* nulls handle */
...
if (!(default_q||fmc_q))            deconfig         /* handle NULL */
                                    deconfig         /* unconditional */

First call closes the port and sets port_handle = NULL; the next calls
hit fm_port_disable(NULL) (and delete_pcd/scheme/netenv on stale
handles). In default_q/fmc_q mode the last call also runs with a
possibly-NULL handle. Restore the !port_handle guard in
dpaa_fm_deconfig(), or guard every call site and drop the duplicates;
there is no reason to deconfig three times. dpaa_port_vsp_cleanup() is
fine, it is idempotent.

Patch 13: bus/dpaa improve log macro and fix bus detection
Error: dpaa_bus_dev_compare() no longer returns a comparison result.
It is installed as bus->dev_compare and rte_bus_find_devargs() uses it
as "cmp(name, devargs_name) != 0 -> continue". The new body returns 0
on any DPAA platform (and short-circuits to 0 once dpaa_bus.detected is
set), so every device name compares "equal" to the first devargs in the
list, and per-device devargs get misapplied. A comparator must return
the name comparison (old strncmp(devname1, devname2, ...)). Bus
detection and pthread_key_create() side effects do not belong in a
compare callback.

Patch 03: drivers add BMI Tx statistics
Warning: the Tx BMI counters are added to dpaa_xstats_strings but
nothing reads them. fman_if_bmi_stats_get_all() and
fman_if_bmi_stats_reset() still only touch the Rx registers via
rx_bmi_map, and FMAN_IF_BMI_TX_STAT_OFFSET_START/END are defined but
unused. In dpaa_dev_xstats_get()/dpaa_xstats_get_by_id() the
regular-vs-BMI split is computed from sizeof(struct
dpaa_if_rx_bmi_stats) only, so adding 4 Tx entries shifts the boundary
and the Tx counters report stale/Rx data. Either wire get_all/reset to
read the tx_bmi_map range and fix the bmi_count split, or drop the Tx
entries from this patch.

Patch 04: drivers add process-type guards for secondary process
Error: dpaa_qdma_init() adds "char *penv;" but does not use it here;
penv is only consumed in patch 18. This commit fails to build
standalone with -Werror (-Wunused-variable), breaking bisect. Move the
declaration to patch 18.
Info: the subject says "add process-type guards" but the dpaa_sec hunk
removes the RTE_PROC_PRIMARY early-return in cryptodev_dpaa_sec_probe().
That is actually correct (secondary still attaches the cryptodev, and
dpaa_sec_dev_init() has its own primary-only guard), but the commit
message should say so.

Patch 18: dma/dpaa add SG data validation and ERR050757 fix
Warning: five new getenv() tunables (DPAA_QDMA_DATA_VALIDATION,
DPAA_QDMA_HW_ERR_CHECK, DPAA_QDMA_SG_ENABLE, DPAA_QDMA_SG_MAX_ENTRY_SIZE,
DPAA_QDMA_PCI_READ). The driver already uses devargs
(DPAA_DMA_ERROR_CHECK); these should be devargs too. checkpatch will
also flag the getenv additions.
Info: s_hw_err_check changes from bool to int; keep bool for a
true/false flag.

Patch 08: bus/dpaa enhance DPAA FQ shutdown
Info: qman_find_fq_by_cgrid() ends with "do {...} while (1);" followed
by an unreachable "return -ENODEV;" - drop the dead return. The linear
FQID probe from 1 with a hardware query per iteration relies on
qman_query_fq_np() returning -ERANGE to terminate; worth a comment.
This patch also leaves a new blank line at EOF (checkpatch).

Patch 17: drivers release DPAA bpid on driver destructor
Info: "#define RTE_PRIORITY_104 104" is unused; RTE_FINI_PRIO is called
with the literal 104. Use the macro or drop it.