mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Tushar Dave <tdave@nvidia.com>
To: Jason Gunthorpe <jgg@nvidia.com>
Cc: corbet@lwn.net, bhelgaas@google.com, paulmck@kernel.org,
	akpm@linux-foundation.org, thuth@redhat.com, rostedt@goodmis.org,
	xiongwei.song@windriver.com, vidyas@nvidia.com,
	linux-doc@vger.kernel.org, linux-kernel@vger.kernel.org,
	linux-pci@vger.kernel.org, vsethi@nvidia.com,
	sdonthineni@nvidia.com
Subject: Re: [PATCH 1/1] PCI: Fix Extend ACS configurability
Date: Mon, 6 Jan 2025 12:34:00 -0800	[thread overview]
Message-ID: <2676cf6e-d9eb-4a34-be5e-29824458f92f@nvidia.com> (raw)
In-Reply-To: <20250102184009.GD5556@nvidia.com>



On 1/2/25 10:40, Jason Gunthorpe wrote:
> On Fri, Dec 13, 2024 at 12:29:42PM -0800, Tushar Dave wrote:
> 
>> diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
>> index dc663c0ca670..fc1c37910d1c 100644
>> --- a/Documentation/admin-guide/kernel-parameters.txt
>> +++ b/Documentation/admin-guide/kernel-parameters.txt
>> @@ -4654,11 +4654,10 @@
>>   				Format:
>>   				<ACS flags>@<pci_dev>[; ...]
>>   				Specify one or more PCI devices (in the format
>> -				specified above) optionally prepended with flags
>> -				and separated by semicolons. The respective
>> -				capabilities will be enabled, disabled or
>> -				unchanged based on what is specified in
>> -				flags.
>> +				specified above) prepended with flags and separated
>> +				by semicolons. The respective capabilities will be
>> +				enabled, disabled or unchanged based on what is
>> +				specified in flags.
>>   
>>   				ACS Flags is defined as follows:
>>   				  bit-0 : ACS Source Validation
>> @@ -4673,7 +4672,7 @@
>>   				  '1' – force enabled
>>   				  'x' – unchanged
>>   				For example,
>> -				  pci=config_acs=10x
>> +				  pci=config_acs=10x@pci:0:0
>>   				would configure all devices that support
>>   				ACS to enable P2P Request Redirect, disable
>>   				Translation Blocking, and leave Source
> 
> Is this an unrelated change? The format of the command line shouldn't
> be changed to fix the described bug, why is the documentation changed?

The documentation as it is (i.e. without my patch), is not correct.

IOW, config_acs parameter does require flags and it is not optional. Without 
flags it results into "ACS Flags missing". Therefore I remove word "optionally" 
from the documentation text.

Secondly, the syntax in the example 'pci=config_acs=10x' is incorrect. The 
correct syntax should be 'pci=config_acs=10x@pci:0:0' that would configure all 
devices that support ACS to enable P2P Request Redirect, disable Translation 
Blocking, and leave Source Validation unchanged from whatever power-up or 
firmware set it to.

> 
>>   static void __pci_config_acs(struct pci_dev *dev, struct pci_acs *caps,
>> -			     const char *p, u16 mask, u16 flags)
>> +			     const char *p, const u16 acs_mask, const u16 acs_flags)
>>   {
>> +	u16 flags = acs_flags;
>> +	u16 mask = acs_mask;
>>   	char *delimit;
>>   	int ret = 0;
>>   
>> @@ -964,7 +965,7 @@ static void __pci_config_acs(struct pci_dev *dev, struct pci_acs *caps,
>>   		return;
>>   
>>   	while (*p) {
>> -		if (!mask) {
>> +		if (!acs_mask) {
>>   			/* Check for ACS flags */
>>   			delimit = strstr(p, "@");
>>   			if (delimit) {
>> @@ -972,6 +973,8 @@ static void __pci_config_acs(struct pci_dev *dev, struct pci_acs *caps,
>>   				u32 shift = 0;
>>   
>>   				end = delimit - p - 1;
>> +				mask = 0;
>> +				flags = 0;
>>   
>>   				while (end > -1) {
>>   					if (*(p + end) == '0') {
> 
> This function the entire fix, right? Because the routine was
> clobbering acs_mask as it processed the earlier devices?

yes, that is correct.

> 
>> @@ -1028,10 +1031,10 @@ static void __pci_config_acs(struct pci_dev *dev, struct pci_acs *caps,
>>   
>>   	pci_dbg(dev, "ACS mask  = %#06x\n", mask);
>>   	pci_dbg(dev, "ACS flags = %#06x\n", flags);
>> +	pci_dbg(dev, "ACS control = %#06x\n", caps->ctrl);
>>   
>> -	/* If mask is 0 then we copy the bit from the firmware setting. */
>> -	caps->ctrl = (caps->ctrl & ~mask) | (caps->fw_ctrl & mask);
>> -	caps->ctrl |= flags;
>> +	caps->ctrl &= ~mask;
>> +	caps->ctrl |= (flags & mask);
> 
> And why delete fw_ctrl? Doesn't that break the unchanged
> functionality?

No, it does not break the unchanged functionality. I removed it because it is 
not needed after my fix.

If it helps, using 'config_acs' the code only allows to configures the lower 7 
bits of ACS ctrl for the specified PCI device(s).
The bits other than the lower 7 bits of ACS ctrl remain unchanged.
The bits specified with 'x' or 'X' that are within the 7 lower bits remain 
unchanged. Trying to configure bits other than lower 7 bits generates an error 
message "Invalid ACS flags specified"

-Tushar
> 
> Jason

  reply	other threads:[~2025-01-06 20:34 UTC|newest]

Thread overview: 18+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-12-13 20:29 Tushar Dave
2024-12-14 12:30 ` Vidya Sagar
2025-01-02 18:40 ` Jason Gunthorpe
2025-01-06 20:34   ` Tushar Dave [this message]
2025-01-06 20:53     ` Bjorn Helgaas
2025-01-08  2:34       ` Tushar Dave
2025-01-07  0:10     ` Jason Gunthorpe
2025-01-08  2:32       ` Tushar Dave
2025-01-08 15:10         ` Jason Gunthorpe
2025-01-09  3:13           ` Tushar Dave
2025-01-13 20:07             ` Jason Gunthorpe
2025-01-16  3:11               ` Tushar Dave
2025-01-16 19:01                 ` Jason Gunthorpe
2025-01-17  1:21                   ` Tushar Dave
2025-01-17 13:28                     ` Jason Gunthorpe
2025-01-17 18:41                       ` Tushar Dave
2025-01-02 23:26 ` Bjorn Helgaas
2025-01-06 20:45   ` Tushar Dave

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=2676cf6e-d9eb-4a34-be5e-29824458f92f@nvidia.com \
    --to=tdave@nvidia.com \
    --cc=akpm@linux-foundation.org \
    --cc=bhelgaas@google.com \
    --cc=corbet@lwn.net \
    --cc=jgg@nvidia.com \
    --cc=linux-doc@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-pci@vger.kernel.org \
    --cc=paulmck@kernel.org \
    --cc=rostedt@goodmis.org \
    --cc=sdonthineni@nvidia.com \
    --cc=thuth@redhat.com \
    --cc=vidyas@nvidia.com \
    --cc=vsethi@nvidia.com \
    --cc=xiongwei.song@windriver.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

all inboxes | Powered by JetHome®