[PATCH] doc: add capability to access physical addresses

[Isaac Boukris]

Hi,

On Sun, Jan 15, 2023 at 2:46 PM Dmitry Kozlyuk <dmitry.kozliuk at gmail.com> wrote:
>
> 2023-01-14 18:27 (UTC-0800), Stephen Hemminger:
> > DAC_OVERRIDE is like having the master key. It opens all doors
> > and if so, running as non-root really doesn't matter that much.
> >
> > Ideally, a finer grain permission could be used.
> > Recommending this to users seems wrong.
>
> According to my tests, DAC_READ_SEARCH can be used instead of DAC_OVERRIDE.
> It seems slightly better, because it doesn't bypass write permission checks.
> Although I agree with Isaac that SYS_ADMIN is already very powerful,
> and remember that the final goal is to perform unrestricted DMA.
> Boris, Isaac, is DAC_READ_SEARCH sufficient on your systems?

Yes, I can confirm that DAC_READ_SEARCH works fine on my system as well. Thanks!

[admin at localhost ~]$ getcap /usr/bin/dpdk-testpmd
/usr/bin/dpdk-testpmd cap_dac_read_search,cap_ipc_lock,cap_sys_admin=ep
[admin at localhost ~]$ dpdk-testpmd -l 2,3,4 -a 000:0b:00.0 --huge-dir
/dev/hugepages/ --iova-mode pa
EAL: Detected CPU lcores: 8
EAL: Detected NUMA nodes: 1
EAL: Detected shared linkage of DPDK
EAL: Multi-process socket /tmp/dpdk/rte/mp_socket
EAL: Selected IOVA mode 'PA'
EAL: No available 1048576 kB hugepages reported
TELEMETRY: No legacy callbacks, legacy socket not created
testpmd: No probed ethernet devices
testpmd: create a new mbuf pool <mb_pool_0>: n=163456, size=2176, socket=0
testpmd: preferred mempool ops selected: ring_mp_mc
Done