From mboxrd@z Thu Jan 1 00:00:00 1970
Return-Path: <dev-bounces@dpdk.org>
Received: from mails.dpdk.org (mails.dpdk.org [217.70.189.124])
by inbox.dpdk.org (Postfix) with ESMTP id 492A0461BE;
Fri, 7 Feb 2025 18:47:10 +0100 (CET)
Received: from mails.dpdk.org (localhost [127.0.0.1])
by mails.dpdk.org (Postfix) with ESMTP id E072742EC7;
Fri, 7 Feb 2025 18:47:09 +0100 (CET)
Received: from linux.microsoft.com (linux.microsoft.com [13.77.154.182])
by mails.dpdk.org (Postfix) with ESMTP id 518E842EC4
for <dev@dpdk.org>; Fri, 7 Feb 2025 18:47:08 +0100 (CET)
Received: by linux.microsoft.com (Postfix, from userid 1213)
id 9E2C72107302; Fri, 7 Feb 2025 09:47:07 -0800 (PST)
DKIM-Filter: OpenDKIM Filter v2.11.0 linux.microsoft.com 9E2C72107302
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=linux.microsoft.com;
s=default; t=1738950427;
bh=jjfsIFLg7z187ZrOTU+2hMBMuyRqLL5sE1ohcdas8wM=;
h=Date:From:To:Cc:Subject:References:In-Reply-To:From;
b=Zi2YMSVlrBl2TtLhBT3S980+5dc05xTn+p4LFQrs64aYZjiUzr5yIjUUuPeXDfCCh
+u78mnpxMcgYWOc8PqD9CeY/mMUTB1FusN7dEXNsDaPI84yfh9FcqGu+WjHFvTos1V
CmZdHochVGp+Xyx8fesbVbdVtsPwMDScWL1LHICE=
Date: Fri, 7 Feb 2025 09:47:07 -0800
From: Andre Muezerie <andremue@linux.microsoft.com>
To: Bruce Richardson <bruce.richardson@intel.com>
Cc: anatoly.burakov@intel.com, dev@dpdk.org, ian.stokes@intel.com,
jochen.behrens@broadcom.com, vladimir.medvedkin@intel.com,
stephen@networkplumber.org
Subject: Re: [PATCH v4] drivers/net: use 64-bit shift and avoid
signed/unsigned mismatch
Message-ID: <20250207174707.GA21139@linuxonhyperv3.guj3yctzbm1etfxqx2vob5hsef.xx.internal.cloudapp.net>
References: <1735246770-731-1-git-send-email-andremue@linux.microsoft.com>
<1738783944-10172-1-git-send-email-andremue@linux.microsoft.com>
<Z6SX9MGQC3f776nN@bricha3-mobl1.ger.corp.intel.com>
<20250207154603.GB21754@linuxonhyperv3.guj3yctzbm1etfxqx2vob5hsef.xx.internal.cloudapp.net>
<Z6YtOvnCZp7Ta7Fq@bricha3-mobl1.ger.corp.intel.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <Z6YtOvnCZp7Ta7Fq@bricha3-mobl1.ger.corp.intel.com>
User-Agent: Mutt/1.5.21 (2010-09-15)
X-BeenThere: dev@dpdk.org
X-Mailman-Version: 2.1.29
Precedence: list
List-Id: DPDK patches and discussions <dev.dpdk.org>
List-Unsubscribe: <https://mails.dpdk.org/options/dev>,
<mailto:dev-request@dpdk.org?subject=unsubscribe>
List-Archive: <http://mails.dpdk.org/archives/dev/>
List-Post: <mailto:dev@dpdk.org>
List-Help: <mailto:dev-request@dpdk.org?subject=help>
List-Subscribe: <https://mails.dpdk.org/listinfo/dev>,
<mailto:dev-request@dpdk.org?subject=subscribe>
Errors-To: dev-bounces@dpdk.org
On Fri, Feb 07, 2025 at 03:56:42PM +0000, Bruce Richardson wrote:
> On Fri, Feb 07, 2025 at 07:46:03AM -0800, Andre Muezerie wrote:
> > On Thu, Feb 06, 2025 at 11:07:32AM +0000, Bruce Richardson wrote:
> > > On Wed, Feb 05, 2025 at 11:32:24AM -0800, Andre Muezerie wrote:
> > > > This patch avoids warnings like the ones below emitted by MSVC:
> > > >
> > > > 1)
> > > > ../drivers/net/ice/base/ice_flg_rd.c(71): warning C4334: '<<':
> > > > result of 32-bit shift implicitly converted to 64 bits
> > > > (was 64-bit shift intended?)
> > > >
> > > > 2)
> > > > ../drivers/net/ice/ice_dcf_sched.c(177): warning C4018: '>=':
> > > > signed/unsigned mismatch
> > > >
> > > > The fix for (1) is to use 64-bit shifting when appropriate
> > > > (according to what the result is used for).
> > > >
> > > > The fix for (2) is to explicitly cast the variables used in the
> > > > comparison.
> > > >
> > > > Signed-off-by: Andre Muezerie <andremue@linux.microsoft.com>
> > > > ---
> > > > drivers/net/intel/i40e/i40e_ethdev.c | 22 +++++++++++-----------
> > > > drivers/net/intel/iavf/iavf_ethdev.c | 2 +-
> > > > drivers/net/intel/iavf/iavf_rxtx.c | 2 +-
> > > > drivers/net/intel/iavf/iavf_vchnl.c | 2 +-
> > > > drivers/net/intel/ice/base/meson.build | 19 +++++++++++++------
> > > > drivers/net/intel/ice/ice_dcf_sched.c | 2 +-
> > > > drivers/net/intel/ice/ice_ethdev.c | 4 ++--
> > > > drivers/net/intel/ice/ice_rxtx.c | 2 +-
> > > > drivers/net/intel/ixgbe/ixgbe_ethdev.c | 2 +-
> > > > drivers/net/vmxnet3/vmxnet3_ethdev.h | 6 +++---
> > > > 10 files changed, 35 insertions(+), 28 deletions(-)
> > > >
>
> <snip>
>
> > > > diff --git a/drivers/net/intel/ice/base/meson.build b/drivers/net/intel/ice/base/meson.build
> > > > index addb922ac9..dc5956f92c 100644
> > > > --- a/drivers/net/intel/ice/base/meson.build
> > > > +++ b/drivers/net/intel/ice/base/meson.build
> > > > @@ -31,18 +31,25 @@ sources = [
> > > > 'ice_vf_mbx.c',
> > > > ]
> > > >
> > > > -error_cflags = [
> > > > - '-Wno-unused-but-set-variable',
> > > > - '-Wno-unused-variable',
> > > > - '-Wno-unused-parameter',
> > > > -]
> > > > +if is_ms_compiler
> > > > + error_cflags = [
> > > > + '/wd4101', # unreferenced local variable
> > > > + '/wd4334', # result of 32-bit shift implicitly converted to 64 bits
> > > > + ]
> > > > +else
> > > > + error_cflags = [
> > > > + '-Wno-unused-but-set-variable',
> > > > + '-Wno-unused-variable',
> > > > + '-Wno-unused-parameter',
> > > > + ]
> > > > +endif
> > > >
> > >
> > > Do we actually need these if-else blocks here? The way
> > > the code is structured is that we check if the flags work to the current
> > > compiler and use only those that are relevant. Therefore, we should just be
> > > able to have a list of error flags and leave meson to filter out the
> > > incorrect ones.
> >
> > Both approaches work. I personally find the if-else approach in this case a little more readable
> > as it makes clear to which compiler the flags apply (considering that there might be multiple
> > flags with the same purpose, one for each compiler). But I'm open to updating the patch following
> > your suggestion.
> >
>
> If you find it more readable, ok to keep as-is.
Ok. I'll keep this as-is then.
>
> > >
> > > > # Bugzilla ID: 678
> > > > if (toolchain == 'gcc' and cc.version().version_compare('>=11.0.0'))
> > > > error_cflags += ['-Wno-array-bounds']
> > > > endif
> > > >
> > > > -if is_windows and cc.get_id() != 'clang'
> > > > +if is_windows and not is_ms_compiler and cc.get_id() != 'clang'
> > >
> > > Are there other supported compiler options for windows other than MSVC and
> > > clang? For what compiler are we adding this flag?
> >
> > Yes, MinGW-w64 is also supported on Windows, so effectively this flag applies to this compiler.
> > https://doc.dpdk.org/guides/windows_gsg/build_dpdk.html
> > Note that this flag was already there. I just changed the expression so that it is not used with msvc.
> >
>
> Ok, thanks for clarifying.
> Do we have a flag or check for identifying MinGW, because if we do that may
> be clearer in the check.
I checked on my system and MinGW identifies as "gcc". We can probably rely on that, so I simplified the expression as suggested.
>
> > >
> > > > cflags += ['-fno-asynchronous-unwind-tables']
> > > > endif
> > > >
> > > > diff --git a/drivers/net/intel/ice/ice_dcf_sched.c b/drivers/net/intel/ice/ice_dcf_sched.c
> > > > index 7967c35533..2832d223d1 100644
> > > > --- a/drivers/net/intel/ice/ice_dcf_sched.c
> > > > +++ b/drivers/net/intel/ice/ice_dcf_sched.c
> > > > @@ -174,7 +174,7 @@ ice_dcf_node_param_check(struct ice_dcf_hw *hw, uint32_t node_id,
> > > > }
> > > >
> > > > /* for non-leaf node */
> > > > - if (node_id >= 8 * hw->num_vfs) {
> > > > + if (node_id >= (uint32_t)(8 * hw->num_vfs)) {
> > > > if (params->nonleaf.wfq_weight_mode) {
> > > > error->type =
> > > > RTE_TM_ERROR_TYPE_NODE_PARAMS_WFQ_WEIGHT_MODE;
> > > > diff --git a/drivers/net/intel/ice/ice_ethdev.c b/drivers/net/intel/ice/ice_ethdev.c
> > > > index 80eee03204..6f6f618a2f 100644
> > > > --- a/drivers/net/intel/ice/ice_ethdev.c
> > > > +++ b/drivers/net/intel/ice/ice_ethdev.c
> > > > @@ -2469,13 +2469,13 @@ ice_get_supported_rxdid(struct ice_hw *hw)
> > > > uint32_t regval;
> > > > int i;
> > > >
> > > > - supported_rxdid |= BIT(ICE_RXDID_LEGACY_1);
> > > > + supported_rxdid |= RTE_BIT64(ICE_RXDID_LEGACY_1);
> > > >
> > > > for (i = ICE_RXDID_FLEX_NIC; i < ICE_FLEX_DESC_RXDID_MAX_NUM; i++) {
> > > > regval = ICE_READ_REG(hw, GLFLXP_RXDID_FLAGS(i, 0));
> > > > if ((regval >> GLFLXP_RXDID_FLAGS_FLEXIFLAG_4N_S)
> > > > & GLFLXP_RXDID_FLAGS_FLEXIFLAG_4N_M)
> > > > - supported_rxdid |= BIT(i);
> > > > + supported_rxdid |= RTE_BIT64(i);
> > > > }
> > > > return supported_rxdid;
> > > > }
> > > > diff --git a/drivers/net/intel/ice/ice_rxtx.c b/drivers/net/intel/ice/ice_rxtx.c
> > > > index 8dd8644b16..87a9d93e89 100644
> > > > --- a/drivers/net/intel/ice/ice_rxtx.c
> > > > +++ b/drivers/net/intel/ice/ice_rxtx.c
> > > > @@ -399,7 +399,7 @@ ice_program_hw_rx_queue(struct ice_rx_queue *rxq)
> > > > PMD_DRV_LOG(DEBUG, "Port (%u) - Rx queue (%u) is set with RXDID : %u",
> > > > rxq->port_id, rxq->queue_id, rxdid);
> > > >
> > > > - if (!(pf->supported_rxdid & BIT(rxdid))) {
> > > > + if (!(pf->supported_rxdid & RTE_BIT64(rxdid))) {
> > > > PMD_DRV_LOG(ERR, "currently package doesn't support RXDID (%u)",
> > > > rxdid);
> > > > return -EINVAL;
> > > > diff --git a/drivers/net/intel/ixgbe/ixgbe_ethdev.c b/drivers/net/intel/ixgbe/ixgbe_ethdev.c
> > > > index 5f18fbaad5..078f7b47c3 100644
> > > > --- a/drivers/net/intel/ixgbe/ixgbe_ethdev.c
> > > > +++ b/drivers/net/intel/ixgbe/ixgbe_ethdev.c
> > > > @@ -2722,7 +2722,7 @@ ixgbe_dev_start(struct rte_eth_dev *dev)
> > > > ixgbe_set_vf_rate_limit(
> > > > dev, vf,
> > > > vfinfo[vf].tx_rate[idx],
> > > > - 1 << idx);
> > > > + RTE_BIT64(idx));
> > > > }
> > > >
> > > > ixgbe_restore_statistics_mapping(dev);
> > > > diff --git a/drivers/net/vmxnet3/vmxnet3_ethdev.h b/drivers/net/vmxnet3/vmxnet3_ethdev.h
> > > > index e9ded6663d..e59cb285f4 100644
> > > > --- a/drivers/net/vmxnet3/vmxnet3_ethdev.h
> > > > +++ b/drivers/net/vmxnet3/vmxnet3_ethdev.h
> > > > @@ -186,14 +186,14 @@ static inline uint8_t
> > > > vmxnet3_get_ring_idx(struct vmxnet3_hw *hw, uint32 rqID)
> > > > {
> > > > return (rqID >= hw->num_rx_queues &&
> > > > - rqID < 2 * hw->num_rx_queues) ? 1 : 0;
> > > > + rqID < (uint32)2 * hw->num_rx_queues) ? 1 : 0;
> > >
> > > Why uint32 rather than uint32_t which are the normal types we use in DPDK?
> > > Could this also be simplified to just 2U?
> >
> > Well, I had just used the same type used for rqID (a few lines above).
> > BTW, there are more than 100 hits in DPDK when searching for "uint32 ".
> > I'm happy to use 2U instead though.
> >
>
> I generally don't like to see uint32 rather than the normal uint32_t type.
> Probably prefer 2U though since it's shorter. However, if you don't feel
> like doing a new revision, ok to keep this as you have it.
>
> /Bruce
I replaced that with 2U. I also took the opportunity to update the type in the
function signature and also updated the variable name to avoid a checkpatch
warning.
--
Andre Muezerie