RE: [EXTERNAL] Re: [PATCH v6 7/8] app: reduce app dependencies

[Akhil Goyal <gakhil@marvell.com>]

> On Wed, Feb 19, 2025 at 11:18:46AM +0100, Morten Brørup wrote:
> > > From: Akhil Goyal [mailto:gakhil@marvell.com]
> > > Sent: Wednesday, 19 February 2025 09.16
> > >
> > > Hi David,
> > > > Hello,
> > > >
> > > > Cc: Akhil.
> > > >
> > > > On Tue, Feb 18, 2025 at 12:16 PM Anatoly Burakov
> > > > <anatoly.burakov@intel.com> wrote:
> > > > > diff --git a/app/test-crypto-perf/meson.build b/app/test-crypto-
> > > > perf/meson.build
> > > > > index fb48d9ec29..f056431bf4 100644
> > > > > --- a/app/test-crypto-perf/meson.build
> > > > > +++ b/app/test-crypto-perf/meson.build
> > > > > @@ -19,7 +19,7 @@ sources = files(
> > > > >          'cperf_test_verify.c',
> > > > >          'main.c',
> > > > >  )
> > > > > -deps += ['cryptodev', 'net', 'security']
> > > > > +deps += ['cryptodev']
> > > > >  optional_deps += 'crypto_scheduler'
> > > > >
> > > > >  cflags += no_wvla_cflag
> > > >
> > > > - This app does seem to rely on lib/net.
> > > >
> > > > [989/1266] Compiling C object
> > > > app/dpdk-test-crypto-perf.p/test-crypto-perf_cperf_ops.c.o
> > > > FAILED: app/dpdk-test-crypto-perf.p/test-crypto-perf_cperf_ops.c.o
> > > > ccache aarch64-linux-gnu-gcc -Iapp/dpdk-test-crypto-perf.p -Iapp
> > > > -I../app -Ilib/cryptodev -I../lib/cryptodev -Ilib/mbuf -I../lib/mbuf
> > > > -Ilib/mempool -I../lib/mempool -Ilib/ring -I../lib/ring -I. -I..
> > > > -Iconfig -I../config -Ilib/eal/include -I../lib/eal/include
> > > > -Ilib/eal/linux/include -I../lib/eal/linux/include
> > > > -Ilib/eal/arm/include -I../lib/eal/arm/include -I../kernel/linux
> > > > -Ilib/eal/common -I../lib/eal/common -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/rcu
> > > > -I../lib/rcu -fdiagnostics-color=always -pipe -D_FILE_OFFSET_BITS=64
> > > > -Wall -Winvalid-pch -Wextra -Werror -std=c11 -O3 -include
> > > rte_config.h
> > > > -Wvla -Wcast-qual -Wdeprecated -Wformat -Wformat-nonliteral
> > > > -Wformat-security -Wmissing-declarations -Wmissing-prototypes
> > > > -Wnested-externs -Wold-style-definition -Wpointer-arith -Wsign-
> > > compare
> > > > -Wstrict-prototypes -Wundef -Wwrite-strings -Wno-packed-not-aligned
> > > > -Wno-missing-field-initializers -D_GNU_SOURCE -mcpu=cortex-a72
> > > > -DALLOW_EXPERIMENTAL_API -Wno-vla -MD -MQ
> > > > app/dpdk-test-crypto-perf.p/test-crypto-perf_cperf_ops.c.o -MF
> > > > app/dpdk-test-crypto-perf.p/test-crypto-perf_cperf_ops.c.o.d -o
> > > > app/dpdk-test-crypto-perf.p/test-crypto-perf_cperf_ops.c.o -c
> > > > ../app/test-crypto-perf/cperf_ops.c
> > > > ../app/test-crypto-perf/cperf_ops.c:6:10: fatal error: rte_ether.h:
> > > No
> > > > such file or directory
> > > >     6 | #include <rte_ether.h>
> > > >       |          ^~~~~~~~~~~~~
> > > > compilation terminated.
> > > > ninja: build stopped: subcommand failed.
> > > >
> > > >
> > > > - Btw, rte_security.h is included in app/test-crypto-
> > > perf/cperf_ops.c.
> > > > However, given there is a #ifdef, I suspect the intention was to have
> > > > lib/security as an optional dep.
> > >
> > > Yes it was initially meant to be optional when rte_security was
> > > experimental.
> > > But now, it is kind of mandatory library. I am not sure if it will
> > > compile without rte_security.
> > > We can remove the ifdefs over rte_security and
> > > we should not remove security dependency from crypto-perf.
> > >
> > > -Akhil
> >
> > Not sure what you mean here, but:
> > The "security" library is optional, not always enabled, when building DPDK.
> > Ref: /source/lib/meson.build
> >
> > Not all applications use the security library (or stuff that depends on it), so don't
> make it mandatory!
> >
> I believe he's just referring to making it a mandatory dependency for this
> app, not for all of DPDK. (Or so I hope). Currently it's an optional
> dependency of the app.
>
I was talking about all applications which depend on rte_security.

Previously all applications which were supposed to use rte_security,
we made sure that applications also get compiled without rte_security.
(by skipping the rte_security related cases using ifdefs)
But now, that bifurcation is not maintained and 
we can remove ifdefs from all applications which have rte_security dependency.