Re: [dpdk-dev] [PATCH v2] build: disable armv8 crypto extension

[Yongseok Koh <yskoh@mellanox.com>]

On Fri, May 03, 2019 at 04:10:47PM +0000, Honnappa Nagarahalli wrote:
> > 
> > Hi Yongseok,
> > 	We need to enable 'CONFIG_RTE_LIBRTE_PMD_ARMV8_CRYPTO'
> > (which would require a documentation change in [1]).
> I enabled this and compiled, the compilation fails. Ideally (as discussed in
> other threads), the PMD code itself does not make use of the crypto
> instructions, so we should be able to compile the PMDs. Crypto functionality
> should not be available only if crypto is not available from the CPU or the
> crypto library is not present. Otherwise, there is no way to use crypto in
> distro package without recompiling. We can take this up separately. 

Like you mentioned, the failure isn't due to lack of '+crypto' flag but the
external library.

CONFIG_RTE_LIBRTE_PMD_ARMV8_CRYPTO is disabled by default because it needs the
external library. In order to enable the config, ARMV8_CRYPTO_LIB_PATH must be
specified. Makefile mandates it.

	ifneq ($(MAKECMDGOALS),clean)
	ifneq ($(MAKECMDGOALS),config)
	ifeq ($(ARMV8_CRYPTO_LIB_PATH),)
	$(error "Please define ARMV8_CRYPTO_LIB_PATH environment variable")
	endif
	endif
	endif

armv8 crypto pmd has nothing to do with '+crypto' cpuflag but the external
library (compiled with '+crypto' flag) has to be linked. Without the external
library, it is meaningless to compile the armv8 crypto PMD because we don't have
any fallback.

In runtime, in order to guarantee the final binary runs w/o crash, it checks
cpuflag of the target host with rte_cpu_get_flag_enabled().

>  I think this change might have an impact on the existing users. Does this
>  change need to be documented somewhere (at least in the release notes)?

Practically, there would be no impact. Even if user's app makes use of crypto
instructions, the build config of the app should have the flag.

> > 
> > [1] https://eur03.safelinks.protection.outlook.com/?url=https%3A%2F%2Fdoc.dpdk.org%2Fguides-19.02%2Fcryptodevs%2Farmv8.html&data=02%7C01%7Cyskoh%40mellanox.com%7C4d6fc9819e4e4a5b1ba508d6cfe1e91d%7Ca652971c7d2e4d9ba6a4d149256f461b%7C0%7C0%7C636924966530971290&sdata=r3VFKjX44T1g4ORlibYCuCpCDQl1BPQyhzfEe%2BGz%2Fy8%3D&reserved=0
> > 
> > > -----Original Message-----
> > > From: Yongseok Koh <yskoh@mellanox.com>
> > > Sent: Friday, May 3, 2019 7:28 AM
> > > To: jerinj@marvell.com; thomas@monjalon.net
> > > Cc: dev@dpdk.org; bruce.richardson@intel.com;
> > > pbhagavatula@marvell.com; shahafs@mellanox.com; Gavin Hu (Arm
> > > Technology China) <Gavin.Hu@arm.com>; Honnappa Nagarahalli
> > > <Honnappa.Nagarahalli@arm.com>; stable@dpdk.org
> > > Subject: [PATCH v2] build: disable armv8 crypto extension
> > >
> > > Per armv8 crypto extension support, make build always enable it by
> > > default as long as compiler supports the feature while meson build
> > > only enables it for 'default' machine of generic armv8 architecture.
> > >
> > > It is known that not all the armv8 platforms have the crypto
> > > extension. For example, Mellanox BlueField has a variant which doesn't
> > > have it. If crypto enabled binary runs on such a platform, rte_eal_init() fails.
> > >
> > > '+crypto' flag currently implies only '+aes' and '+sha2' and enabling
> > > it will generate the crypto instructions only when crypto intrinsics are used.
> > > For the devices supporting 8.2 crypto or newer, compiler could
> > > generate such instructions beyond intrinsics or asm code. For example,
> > > compiler can generate 3-way exclusive OR instructions if sha3 is
> > > supported. However, it has to be enabled by adding '+sha3' as of today.
> > >
> > > In DPDK, armv8 cryptodev is the only one which requires the crypto support.
> > > As it even uses external library of Marvell which is compiled out of
> > > DPDK with crypto support and there's run-time check for required
> > > cpuflags, crypto support can be disabled in DPDK.
> > >
> > > Cc: stable@dpdk.org
> > >
> > > Signed-off-by: Yongseok Koh <yskoh@mellanox.com>
> > > ---
> > >
> > > v2:
> > > * disable crypto support instead of having a build config
> > >
> > >  config/arm/meson.build        | 2 +-
> > >  mk/machine/armv8a/rte.vars.mk | 2 +-
> > >  2 files changed, 2 insertions(+), 2 deletions(-)
> > >
> > > diff --git a/config/arm/meson.build b/config/arm/meson.build index
> > > 7fa6ed3105..abc8cf346c 100644
> > > --- a/config/arm/meson.build
> > > +++ b/config/arm/meson.build
> > > @@ -74,7 +74,7 @@ flags_octeontx2_extra = [
> > >  	['RTE_USE_C11_MEM_MODEL', true]]
> > >
> > >  machine_args_generic = [
> > > -	['default', ['-march=armv8-a+crc+crypto']],
> > > +	['default', ['-march=armv8-a+crc']],
> > IIRC, this would impact distro packaging as well. Adding Luca.
> > 
> > >  	['native', ['-march=native']],
> > >  	['0xd03', ['-mcpu=cortex-a53']],
> > >  	['0xd04', ['-mcpu=cortex-a35']],
> > > diff --git a/mk/machine/armv8a/rte.vars.mk
> > > b/mk/machine/armv8a/rte.vars.mk index 8252efbb7b..5e3ffc3adf 100644
> > > --- a/mk/machine/armv8a/rte.vars.mk
> > > +++ b/mk/machine/armv8a/rte.vars.mk
> > > @@ -28,4 +28,4 @@
> > >  # CPU_LDFLAGS =
> > >  # CPU_ASFLAGS =
> > >
> > > -MACHINE_CFLAGS += -march=armv8-a+crc+crypto
> > > +MACHINE_CFLAGS += -march=armv8-a+crc
> > > --
> > > 2.11.0
>