From: Matthias Kaehlcke <mka@chromium.org>
To: Jakub Kicinski <jakub.kicinski@netronome.com>
Cc: Manoj Gupta <manojgupta@chromium.org>,
"David S . Miller" <davem@davemloft.net>,
Simon Horman <simon.horman@netronome.com>,
Dirk van der Merwe <dirk.vandermerwe@netronome.com>,
oss-drivers@netronome.com, netdev@vger.kernel.org,
linux-kernel@vger.kernel.org,
Renato Golin <renato.golin@linaro.org>,
Guenter Roeck <groeck@chromium.org>,
Doug Anderson <dianders@chromium.org>
Subject: Re: [PATCH] nfp: convert nfp_eth_set_bit_config() into a macro
Date: Wed, 4 Oct 2017 11:44:59 -0700 [thread overview]
Message-ID: <20171004184459.GN173745@google.com> (raw)
In-Reply-To: <20171004111724.29a4d7ff@cakuba.netronome.com>
El Wed, Oct 04, 2017 at 11:17:24AM -0700 Jakub Kicinski ha dit:
> On Wed, 4 Oct 2017 10:42:42 -0700, Manoj Gupta wrote:
> > Hi Jakub,
> >
> > I had discussed about supporting this code with some clang developers.
> > However, the consensus was this code relies on a specific GCC optimizer
> > behavior and Clang does not share the same behavior by design.
>
> Hm. I find surprising that constant propagation to inlined functions
> is considered something GCC-specific rather than obvious/basic.
>
> > Note that even GCC can't compile this code at -O0.
>
> Yes, that makes me feel uncomfortable... Could you try this?
The kernel as a whole does not build with -O0, so I couldn't try
that. I know Manoj (who isn't a kernel dev) isolated the
compiletime_assert macros and encountered the same 'undefined
reference' errors with gcc -O0 that we are seeing with clang.
This is due to:
"if you use it in an inlined function and pass an argument of the
function as the argument to the built-in, GCC never returns 1 when
you call the inline function with a string constant or compound
literal (see Compound Literals) and does not return 1 when you pass a
constant numeric value to the inline function unless you specify the
-O option."
https://gcc.gnu.org/onlinedocs/gcc-7.2.0/gcc/Other-Builtins.html
> diff --git a/drivers/net/ethernet/netronome/nfp/nfpcore/nfp_nsp_eth.c b/drivers/net/ethernet/netronome/nfp/nfpcore/nfp_nsp_eth.c
> index f6f7c085f8e0..47251396fcae 100644
> --- a/drivers/net/ethernet/netronome/nfp/nfpcore/nfp_nsp_eth.c
> +++ b/drivers/net/ethernet/netronome/nfp/nfpcore/nfp_nsp_eth.c
> @@ -469,10 +469,10 @@ int nfp_eth_set_configured(struct nfp_cpp *cpp, unsigned int idx, bool configed)
> return nfp_eth_config_commit_end(nsp);
> }
>
> -/* Force inline, FIELD_* macroes require masks to be compilation-time known */
> -static __always_inline int
> +static int
> nfp_eth_set_bit_config(struct nfp_nsp *nsp, unsigned int raw_idx,
> - const u64 mask, unsigned int val, const u64 ctrl_bit)
> + const u64 mask, const unsigned int shift,
> + unsigned int val, const u64 ctrl_bit)
> {
> union eth_table_entry *entries = nfp_nsp_config_entries(nsp);
> unsigned int idx = nfp_nsp_config_idx(nsp);
> @@ -489,11 +489,11 @@ nfp_eth_set_bit_config(struct nfp_nsp *nsp, unsigned int raw_idx,
>
> /* Check if we are already in requested state */
> reg = le64_to_cpu(entries[idx].raw[raw_idx]);
> - if (val == FIELD_GET(mask, reg))
> + if (val == (reg & mask) >> shift)
> return 0;
>
> reg &= ~mask;
> - reg |= FIELD_PREP(mask, val);
> + reg |= (val << shift) & mask;
> entries[idx].raw[raw_idx] = cpu_to_le64(reg);
>
> entries[idx].control |= cpu_to_le64(ctrl_bit);
> @@ -503,6 +503,13 @@ nfp_eth_set_bit_config(struct nfp_nsp *nsp, unsigned int raw_idx,
> return 0;
> }
>
> +#define NFP_ETH_SET_BIT_CONFIG(nsp, raw_idx, mask, val, ctrl_bit) \
> + ({ \
> + __BF_FIELD_CHECK(mask, 0ULL, val, "NFP_ETH_SET_BIT_CONFIG: "); \
> + nfp_eth_set_bit_config(nsp, raw_idx, mask, __bf_shf(mask), \
> + val, ctrl_bit); \
> + })
> +
> /**
> * __nfp_eth_set_aneg() - set PHY autonegotiation control bit
> * @nsp: NFP NSP handle returned from nfp_eth_config_start()
> @@ -515,7 +522,7 @@ nfp_eth_set_bit_config(struct nfp_nsp *nsp, unsigned int raw_idx,
> */
> int __nfp_eth_set_aneg(struct nfp_nsp *nsp, enum nfp_eth_aneg mode)
> {
> - return nfp_eth_set_bit_config(nsp, NSP_ETH_RAW_STATE,
> + return NFP_ETH_SET_BIT_CONFIG(nsp, NSP_ETH_RAW_STATE,
> NSP_ETH_STATE_ANEG, mode,
> NSP_ETH_CTRL_SET_ANEG);
> }
> @@ -544,7 +551,7 @@ int __nfp_eth_set_speed(struct nfp_nsp *nsp, unsigned int speed)
> return -EINVAL;
> }
>
> - return nfp_eth_set_bit_config(nsp, NSP_ETH_RAW_STATE,
> + return NFP_ETH_SET_BIT_CONFIG(nsp, NSP_ETH_RAW_STATE,
> NSP_ETH_STATE_RATE, rate,
> NSP_ETH_CTRL_SET_RATE);
> }
> @@ -561,6 +568,6 @@ int __nfp_eth_set_speed(struct nfp_nsp *nsp, unsigned int speed)
> */
> int __nfp_eth_set_split(struct nfp_nsp *nsp, unsigned int lanes)
> {
> - return nfp_eth_set_bit_config(nsp, NSP_ETH_RAW_PORT, NSP_ETH_PORT_LANES,
> + return NFP_ETH_SET_BIT_CONFIG(nsp, NSP_ETH_RAW_PORT, NSP_ETH_PORT_LANES,
> lanes, NSP_ETH_CTRL_SET_LANES);
> }
next prev parent reply other threads:[~2017-10-04 18:44 UTC|newest]
Thread overview: 20+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-10-03 20:05 [PATCH] nfp: convert nfp_eth_set_bit_config() into a macro Matthias Kaehlcke
2017-10-03 21:50 ` Jakub Kicinski
2017-10-04 17:42 ` Matthias Kaehlcke
2017-10-04 17:44 ` David Miller
[not found] ` <CAAMbb05G=HBQweiWqYva_9zTnQqAcwMhJ0yYBUi26T04YA4CxQ@mail.gmail.com>
2017-10-04 18:17 ` Jakub Kicinski
2017-10-04 18:44 ` Matthias Kaehlcke [this message]
2017-10-24 16:56 ` Matthias Kaehlcke
2017-10-24 17:03 ` Jakub Kicinski
2017-10-24 17:13 ` Matthias Kaehlcke
2017-10-04 18:07 ` Joe Perches
2017-10-04 18:49 ` Matthias Kaehlcke
2017-10-04 22:22 ` Jakub Kicinski
2017-10-04 23:16 ` Matthias Kaehlcke
2017-10-04 23:25 ` Jakub Kicinski
2017-10-05 0:38 ` Manoj Gupta
2017-10-05 0:56 ` Jakub Kicinski
2017-10-05 1:50 ` Manoj Gupta
2017-10-05 2:06 ` Jakub Kicinski
2017-10-05 2:13 ` Manoj Gupta
2017-10-09 17:29 ` Matthias Kaehlcke
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20171004184459.GN173745@google.com \
--to=mka@chromium.org \
--cc=davem@davemloft.net \
--cc=dianders@chromium.org \
--cc=dirk.vandermerwe@netronome.com \
--cc=groeck@chromium.org \
--cc=jakub.kicinski@netronome.com \
--cc=linux-kernel@vger.kernel.org \
--cc=manojgupta@chromium.org \
--cc=netdev@vger.kernel.org \
--cc=oss-drivers@netronome.com \
--cc=renato.golin@linaro.org \
--cc=simon.horman@netronome.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).