Re: [dpdk-dev] [PATCH v4 1/2] version: 20.11-rc0

["Kinsella, Ray" <mdr@ashroe.eu>]

Hi Thomas,

On 11/08/2020 12:27, Thomas Monjalon wrote:
> Start a new release cycle with empty release notes.
> 
> The ABI version becomes 21.0.
> The ABI major is back to normal, having only one number (21 vs 20.0).
> The map files are updated to the new ABI major number (21).
> The ABI exceptions specific to previous major version are dropped.
> Travis ABI check is disabled because compatibility is not preserved.
> 
> Signed-off-by: Thomas Monjalon <thomas@monjalon.net>
> ---
> v4: add -rc0 and remove temporary ABI exceptions
> v3: add minor number to ABI_VERSION and remove workaround in scripts
> v2: bump major ABI number (including .map changes)
> ---
>  .travis.yml                                   |  26 ----
>  ABI_VERSION                                   |   2 +-
>  VERSION                                       |   2 +-
>  config/meson.build                            |   5 +-
>  devtools/libabigail.abignore                  |  60 --------

[SNIP]

> diff --git a/devtools/libabigail.abignore b/devtools/libabigail.abignore
> index 9fea822f36..2827d5ff7e 100644
> --- a/devtools/libabigail.abignore
> +++ b/devtools/libabigail.abignore
> @@ -8,12 +8,6 @@
>  [suppress_variable]
>          symbol_version = INTERNAL
>  
> -; Ignore ABI 20.0.1 replaced with ABI 21
> -[suppress_function]
> -        symbol_version = DPDK_20.0.1
> -[suppress_variable]
> -        symbol_version = DPDK_20.0.1
> -
>  ; Explicit ignore for driver-only ABI
>  [suppress_type]
>          name = rte_cryptodev_ops


I think we might be safer doing a complete reset of libabigail.ignore.
We should leave only the suppressions for internal and experimental. 

My concern is that if we have suppressions that live in there forever,
it will becomes stale, and eventually we will end up suppressing things we don't intend.

So yes, if we strip out some of these - especially the enumeration suppressions,
We will end up having to suppress them again as we extend/modify the enumerations etc,
but for me that preferable to the risk of unintended supressions. 

What do others think?


> @@ -56,57 +50,3 @@
>  [suppress_type]
>          name = rte_mbuf_ext_shared_info
>          has_data_member_inserted_between = {offset_of(refcnt_atomic), offset_of(refcnt_atomic)}
> -
> -;;;;;;;;;;;;;;;;;;;;;;
> -; Temporary exceptions till DPDK 20.11
> -;;;;;;;;;;;;;;;;;;;;;;
> -; Ignore moving OCTEONTX2 stable functions to INTERNAL
> -[suppress_file]
> -	file_name_regexp = ^librte_common_octeontx2\.
> -[suppress_file]
> -	file_name_regexp = ^librte_mempool_octeontx2\.
> -; Ignore moving mlx5 stable functions to INTERNAL
> -[suppress_file]
> -	file_name_regexp = ^librte_common_mlx5\.
> -; Ignore moving DPAAx stable functions to INTERNAL
> -[suppress_file]
> -        file_name_regexp = ^librte_common_dpaax\.
> -[suppress_file]
> -        file_name_regexp = ^librte_bus_fslmc\.
> -[suppress_file]
> -        file_name_regexp = ^librte_bus_dpaa\.
> -[suppress_variable]
> -        name = rte_dpaa_bpid_info
> -[suppress_variable]
> -        name = rte_dpaa_memsegs
> -[suppress_variable]
> -        name = rte_dpaa2_bpid_info
> -[suppress_function]
> -        name = rte_dpaa2_mbuf_alloc_bulk
> -[suppress_function]
> -        name_regexp = ^dpaa2?_.*tach$
> -; Ignore soname changes for experimental libraries
> -[suppress_file]
> -	file_name_regexp = ^librte_bbdev\.
> -[suppress_file]
> -	file_name_regexp = ^librte_bpf\.
> -[suppress_file]
> -	file_name_regexp = ^librte_compressdev\.
> -[suppress_file]
> -	file_name_regexp = ^librte_fib\.
> -[suppress_file]
> -	file_name_regexp = ^librte_flow_classify\.
> -[suppress_file]
> -	file_name_regexp = ^librte_graph\.
> -[suppress_file]
> -	file_name_regexp = ^librte_ipsec\.
> -[suppress_file]
> -	file_name_regexp = ^librte_node\.
> -[suppress_file]
> -	file_name_regexp = ^librte_rcu\.
> -[suppress_file]
> -	file_name_regexp = ^librte_rib\.
> -[suppress_file]
> -	file_name_regexp = ^librte_telemetry\.
> -[suppress_file]
> -	file_name_regexp = ^librte_stack\.
[SNIP]