linux-kernel - Re: [PATCH 06/10] PCI/DPC: Use FIELD

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <524d1789-6b57-aae-17b7-bf7ae49719dc@linux.intel.com>
Date:   Mon, 16 Oct 2023 18:10:39 +0300 (EEST)
From:   Ilpo Järvinen <ilpo.jarvinen@...ux.intel.com>
To:     Bjorn Helgaas <helgaas@...nel.org>
cc:     linux-pci@...r.kernel.org,
        Jonathan Cameron <Jonathan.Cameron@...wei.com>,
        Krzysztof Wilczyński <kw@...ux.com>,
        Lorenzo Pieralisi <lpieralisi@...nel.org>,
        LKML <linux-kernel@...r.kernel.org>,
        Bjorn Helgaas <bhelgaas@...gle.com>
Subject: Re: [PATCH 06/10] PCI/DPC: Use FIELD_GET()

On Mon, 16 Oct 2023, Ilpo Järvinen wrote:

> On Fri, 13 Oct 2023, Bjorn Helgaas wrote:
> 
> > On Wed, Oct 11, 2023 at 02:01:13PM +0300, Ilpo Järvinen wrote:
> > > On Tue, 10 Oct 2023, Bjorn Helgaas wrote:
> > > > From: Bjorn Helgaas <bhelgaas@...gle.com>
> > > > 
> > > > Use FIELD_GET() to remove dependences on the field position, i.e., the
> > > > shift value.  No functional change intended.
> > > > 
> > > > Signed-off-by: Bjorn Helgaas <bhelgaas@...gle.com>
> > > > ---
> > > >  drivers/pci/pcie/dpc.c        | 9 +++++----
> > > >  drivers/pci/quirks.c          | 2 +-
> > > >  include/uapi/linux/pci_regs.h | 1 +
> > > >  3 files changed, 7 insertions(+), 5 deletions(-)
> > > > 
> > > > diff --git a/drivers/pci/pcie/dpc.c b/drivers/pci/pcie/dpc.c
> > > > index 3ceed8e3de41..6e551f34ec63 100644
> > > > --- a/drivers/pci/pcie/dpc.c
> > > > +++ b/drivers/pci/pcie/dpc.c
> > > > @@ -8,6 +8,7 @@
> > > >  
> > > >  #define dev_fmt(fmt) "DPC: " fmt
> > > >  
> > > > +#include <linux/bitfield.h>
> > > >  #include <linux/aer.h>
> > > >  #include <linux/delay.h>
> > > >  #include <linux/interrupt.h>
> > > > @@ -202,7 +203,7 @@ static void dpc_process_rp_pio_error(struct pci_dev *pdev)
> > > >  
> > > >  	/* Get First Error Pointer */
> > > >  	pci_read_config_word(pdev, cap + PCI_EXP_DPC_STATUS, &dpc_status);
> > > > -	first_error = (dpc_status & 0x1f00) >> 8;
> > > > +	first_error = FIELD_GET(PCI_EXP_DPC_STATUS_FIRST_ERR, dpc_status);
> > > >  
> > > >  	for (i = 0; i < ARRAY_SIZE(rp_pio_error_string); i++) {
> > > >  		if ((status & ~mask) & (1 << i))
> > > > @@ -270,8 +271,8 @@ void dpc_process_error(struct pci_dev *pdev)
> > > >  	pci_info(pdev, "containment event, status:%#06x source:%#06x\n",
> > > >  		 status, source);
> > > >  
> > > > -	reason = (status & PCI_EXP_DPC_STATUS_TRIGGER_RSN) >> 1;
> > > > -	ext_reason = (status & PCI_EXP_DPC_STATUS_TRIGGER_RSN_EXT) >> 5;
> > > > +	reason = FIELD_GET(PCI_EXP_DPC_STATUS_TRIGGER_RSN, status);
> > > > +	ext_reason = FIELD_GET(PCI_EXP_DPC_STATUS_TRIGGER_RSN_EXT, status);
> > > >  	pci_warn(pdev, "%s detected\n",
> > > >  		 (reason == 0) ? "unmasked uncorrectable error" :
> > > >  		 (reason == 1) ? "ERR_NONFATAL" :
> > > 
> > > BTW, it seems we're doing overlapping work here with many of these 
> > > patches. It takes some time to think these through one by one, I don't 
> > > just autorun through them with coccinelle so I've not posted my changes
> > > yet.
> > >
> > > I went to a different direction here and named all the reasons too with 
> > > defines and used & to get the reason in order to be able to compare with 
> > > the named reasons.
> > > 
> > > You also missed convering one 0xfff4 to use define (although I suspect it 
> > > never was your goal to go beyond FIELD_GET() here).
> > 
> > Pure FIELD_GET() and FIELD_PREP() was my goal.
> > 
> > If you have patches you prefer, I'll drop mine.  I did these about a
> > year ago and it seemed like the time to do something with them since
> > you did the PCI_EXP_LNKSTA_NLW ones and to try to prevent overlapping
> > work.  Since we've started, I'd like to get as much of it done this
> > cycle as possible.
> 
> Okay, I suggest you keep your FIELD_GET/PREP() patch since mine is getting 
> more and more complicated. I can build a nice set of small changes about 
> what remains to do in DPC on top of your patch.

Err, actually, there's still the naming of the define, should _FEP be used 
for First Error Pointer for consistency? You should make that small change 
into your patch if you think _FEP is better because of consistency.


-- 
 i.