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: <aec23bb79b9ff7dd7f13eb67460e0605eac22912.camel@redhat.com>
Date: Thu, 24 Oct 2024 10:02:59 +0200
From: Philipp Stanner <pstanner@...hat.com>
To: Takashi Iwai <tiwai@...e.de>
Cc: Damien Le Moal <dlemoal@...nel.org>, Niklas Cassel <cassel@...nel.org>, 
 Sergey Shtylyov <s.shtylyov@....ru>, Basavaraj Natikar
 <basavaraj.natikar@....com>, Jiri Kosina <jikos@...nel.org>,  Benjamin
 Tissoires <bentiss@...nel.org>, Arnd Bergmann <arnd@...db.de>, Greg
 Kroah-Hartman <gregkh@...uxfoundation.org>, Alex Dubov <oakad@...oo.com>,
 Sudarsana Kalluru <skalluru@...vell.com>, Manish Chopra
 <manishc@...vell.com>, "David S. Miller" <davem@...emloft.net>, Eric
 Dumazet <edumazet@...gle.com>, Jakub Kicinski <kuba@...nel.org>, Paolo
 Abeni <pabeni@...hat.com>, Rasesh Mody <rmody@...vell.com>,
 GR-Linux-NIC-Dev@...vell.com, Igor Mitsyanko <imitsyanko@...ntenna.com>,
 Sergey Matyukevich <geomatsi@...il.com>, Kalle Valo <kvalo@...nel.org>,
 Sanjay R Mehta <sanju.mehta@....com>, Shyam Sundar S K
 <Shyam-sundar.S-k@....com>, Jon Mason <jdmason@...zu.us>, Dave Jiang
 <dave.jiang@...el.com>, Allen Hubbe <allenbh@...il.com>, Bjorn Helgaas
 <bhelgaas@...gle.com>, Alex Williamson <alex.williamson@...hat.com>,
 Juergen Gross <jgross@...e.com>, Stefano Stabellini
 <sstabellini@...nel.org>, Oleksandr Tyshchenko
 <oleksandr_tyshchenko@...m.com>, Jaroslav Kysela <perex@...ex.cz>, Takashi
 Iwai <tiwai@...e.com>, Chen Ni <nichen@...as.ac.cn>, Mario Limonciello
 <mario.limonciello@....com>, Ricky Wu <ricky_wu@...ltek.com>, Al Viro
 <viro@...iv.linux.org.uk>, Breno Leitao <leitao@...ian.org>, Kevin Tian
 <kevin.tian@...el.com>, Thomas Gleixner <tglx@...utronix.de>, Ilpo
 Järvinen <ilpo.jarvinen@...ux.intel.com>, Andy
 Shevchenko <andriy.shevchenko@...ux.intel.com>, Mostafa Saleh
 <smostafa@...gle.com>, Jason Gunthorpe <jgg@...pe.ca>, Yi Liu
 <yi.l.liu@...el.com>,  Christian Brauner <brauner@...nel.org>, Ankit
 Agrawal <ankita@...dia.com>, Eric Auger <eric.auger@...hat.com>, Reinette
 Chatre <reinette.chatre@...el.com>, Ye Bin <yebin10@...wei.com>, Marek
 Marczykowski-Górecki <marmarek@...isiblethingslab.com>,
 Pierre-Louis Bossart <pierre-louis.bossart@...ux.dev>, Peter Ujfalusi
 <peter.ujfalusi@...ux.intel.com>, Maarten Lankhorst
 <maarten.lankhorst@...ux.intel.com>, Kai Vehmanen
 <kai.vehmanen@...ux.intel.com>,  Rui Salvaterra <rsalvaterra@...il.com>,
 linux-ide@...r.kernel.org, linux-kernel@...r.kernel.org, 
 linux-input@...r.kernel.org, netdev@...r.kernel.org, 
 linux-wireless@...r.kernel.org, ntb@...ts.linux.dev,
 linux-pci@...r.kernel.org,  kvm@...r.kernel.org,
 xen-devel@...ts.xenproject.org, linux-sound@...r.kernel.org
Subject: Re: [PATCH 02/13] ALSA: hda_intel: Use always-managed version of
 pcim_intx()

On Wed, 2024-10-23 at 17:03 +0200, Takashi Iwai wrote:
> On Wed, 23 Oct 2024 15:50:09 +0200,
> Philipp Stanner wrote:
> > 
> > On Tue, 2024-10-22 at 16:08 +0200, Takashi Iwai wrote:
> > > On Tue, 15 Oct 2024 20:51:12 +0200,
> > > Philipp Stanner wrote:
> > > > 
> > > > pci_intx() is a hybrid function which can sometimes be managed
> > > > through
> > > > devres. To remove this hybrid nature from pci_intx(), it is
> > > > necessary to
> > > > port users to either an always-managed or a never-managed
> > > > version.
> > > > 
> > > > hda_intel enables its PCI-Device with pcim_enable_device().
> > > > Thus,
> > > > it needs
> > > > the always-managed version.
> > > > 
> > > > Replace pci_intx() with pcim_intx().
> > > > 
> > > > Signed-off-by: Philipp Stanner <pstanner@...hat.com>
> > > > ---
> > > >  sound/pci/hda/hda_intel.c | 2 +-
> > > >  1 file changed, 1 insertion(+), 1 deletion(-)
> > > > 
> > > > diff --git a/sound/pci/hda/hda_intel.c
> > > > b/sound/pci/hda/hda_intel.c
> > > > index b4540c5cd2a6..b44ca7b6e54f 100644
> > > > --- a/sound/pci/hda/hda_intel.c
> > > > +++ b/sound/pci/hda/hda_intel.c
> > > > @@ -786,7 +786,7 @@ static int azx_acquire_irq(struct azx
> > > > *chip,
> > > > int do_disconnect)
> > > >  	}
> > > >  	bus->irq = chip->pci->irq;
> > > >  	chip->card->sync_irq = bus->irq;
> > > > -	pci_intx(chip->pci, !chip->msi);
> > > > +	pcim_intx(chip->pci, !chip->msi);
> > > >  	return 0;
> > > >  }
> > > >  
> > > 
> > > Hm, it's OK-ish to do this as it's practically same as what
> > > pci_intx()
> > > currently does.  But, the current code can be a bit inconsistent
> > > about
> > > the original intx value.  pcim_intx() always stores !enable to
> > > res->orig_intx unconditionally, and it means that the orig_intx
> > > value
> > > gets overridden at each time pcim_intx() gets called.
> > 
> > Yes.
> > 
> > > 
> > > Meanwhile, HD-audio driver does release and re-acquire the
> > > interrupt
> > > after disabling MSI when something goes wrong, and pci_intx()
> > > call
> > > above is a part of that procedure.  So, it can rewrite the
> > > res->orig_intx to another value by retry without MSI.  And after
> > > the
> > > driver removal, it'll lead to another state.
> > 
> > I'm not sure that I understand this paragraph completely. Still,
> > could
> > a solution for the driver on the long-term just be to use
> > pci_intx()?
> 
> pci_intx() misses the restore of the original value, so it's no
> long-term solution, either.

Sure that is missing – I was basically asking whether the driver could
live without that feature.

Consider that point obsolete, see below

> 
> What I meant is that pcim_intx() blindly assumes the negative of the
> passed argument as the original state, which isn't always true.  e.g.
> when the driver calls it twice with different values, a wrong value
> may be remembered.

Ah, I see – thoguh the issue is when it's called several times with the
*same* value, isn't it?

E.g.

pcim_intx(pdev, 1); // 0 is remembered as the old value
pcim_intx(pdev, 1); // 0 is falsely remembered as the old value

Also, it would seem that calling the function for the first time like
that:

pcim_intx(pdev, 0); // old value: 1

is at least incorrect, because INTx should be 0 per default, shouldn't
it? Could then even be a 1st class bug, because INTx would end up being
enabled despite having been disabled all the time.

> 
> That said, I thought of something like below.

At first glance that looks like a good idea to me, thanks for working
this out!

IMO you can submit that as a patch so we can discuss it separately.

Greetings,
Philipp

> 
> 
> thanks,
> 
> Takashi
> 
> -- 8< --
> --- a/drivers/pci/devres.c
> +++ b/drivers/pci/devres.c
> @@ -438,8 +438,17 @@ static void pcim_intx_restore(struct device
> *dev, void *data)
>  	__pcim_intx(pdev, res->orig_intx);
>  }
>  
> -static struct pcim_intx_devres *get_or_create_intx_devres(struct
> device *dev)
> +static void save_orig_intx(struct pci_dev *pdev)
>  {
> +	u16 pci_command;
> +
> +	pci_read_config_word(pdev, PCI_COMMAND, &pci_command);
> +	res->orig_intx = !(pci_command & PCI_COMMAND_INTX_DISABLE);
> +}
> +
> +static struct pcim_intx_devres *get_or_create_intx_devres(struct
> pci_dev *pdev)
> +{
> +	struct device *dev = &pdev->dev;
>  	struct pcim_intx_devres *res;
>  
>  	res = devres_find(dev, pcim_intx_restore, NULL, NULL);
> @@ -447,8 +456,10 @@ static struct pcim_intx_devres
> *get_or_create_intx_devres(struct device *dev)
>  		return res;
>  
>  	res = devres_alloc(pcim_intx_restore, sizeof(*res),
> GFP_KERNEL);
> -	if (res)
> +	if (res) {
> +		save_orig_intx(pdev);
>  		devres_add(dev, res);
> +	}
>  
>  	return res;
>  }
> @@ -467,11 +478,10 @@ int pcim_intx(struct pci_dev *pdev, int enable)
>  {
>  	struct pcim_intx_devres *res;
>  
> -	res = get_or_create_intx_devres(&pdev->dev);
> +	res = get_or_create_intx_devres(pdev);
>  	if (!res)
>  		return -ENOMEM;
>  
> -	res->orig_intx = !enable;
>  	__pcim_intx(pdev, enable);
>  
>  	return 0;
> 


Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ