Received-SPF: pass (google.com: domain of linux-kernel+bounces-65432-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.199.223 as permitted sender) client-ip=147.75.199.223;
Message-ID: <1a46d8cb-104f-4854-a09e-c60095e2dcd0@amd.com>
Date: Wed, 14 Feb 2024 10:33:15 -0500
User-Agent: Mozilla Thunderbird
Cc: yazen.ghannam@amd.com, tony.luck@intel.com, linux-edac@vger.kernel.org,
 linux-kernel@vger.kernel.org, avadhut.naik@amd.com, john.allen@amd.com,
 muralidhara.mk@amd.com, naveenkrishna.chatradhi@amd.com,
 sathyapriya.k@amd.com
Subject: Re: [PATCH 2/2] RAS: Introduce the FRU Memory Poison Manager
Content-Language: en-US
To: Borislav Petkov <bp@alien8.de>
References: <20240214033516.1344948-1-yazen.ghannam@amd.com>
 <20240214033516.1344948-3-yazen.ghannam@amd.com>
 <20240214120214.GJZcyrxgyLLwQ8y19Z@fat_crate.local>
From: Yazen Ghannam <yazen.ghannam@amd.com>
In-Reply-To: <20240214120214.GJZcyrxgyLLwQ8y19Z@fat_crate.local>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit
Precedence: bulk
MIME-Version: 1.0
X-MS-Exchange-AntiSpam-MessageData-ChunkCount: 1
X-MS-Exchange-AntiSpam-MessageData-0:
	=?utf-8?B?ZFZWRS9MbTE1OWNJOUVBVGZIdThKbXNNb2RlMFNtRmtxeDVmaDEwN2tYenZL?=
 =?utf-8?B?dFlEU28wcWpkSGQydHcyQnErcjJjQkg1emRqOXBMM00wZWpuS1RkakdNZkFK?=
 =?utf-8?B?OHR0LzNvRW1WVkdNVXFGbVFmNEtUVHZkUkxSK3lraWkveEdnNnhYQVRFVmRo?=
 =?utf-8?B?M2szd1FJOG5uTDg5MFpXYTAzWEYwaXRvRGpxNGhEV082YXFGSVVWSGxzcFY1?=
 =?utf-8?B?aHpsT0Q0cWRjR1FGT0I2TFlqV2YreW8yVVNFNC9zWDZBdkFiWEhEaTQ0VSsx?=
 =?utf-8?B?RDAxa2ptOVpZSTN6dTFDZzc2ZzdPUnlDSFNqcm1DRmRBZGM5N1JsTHlONXJ5?=
 =?utf-8?B?NktnWEZFSGRRZk1PbHFkaFZjS0lRbno0S1VtOSt6RytNRzlyaHJBVXVGZURG?=
 =?utf-8?B?ZlpvQzg1MGkwN3Rua0NzeXE2NitwcHRYdWIyeTFGQWNOV3FFZXhTemhvckhP?=
 =?utf-8?B?NFZCeXVSWU1KOU5UZHZsSmRPMWhGOStqdzhBV0JDZ09kMmRZeUU0LzMzOFI5?=
 =?utf-8?B?UnVSeVVHdnNwSFpFYVF1dlZaOEVxYlFEY09UR0pFV2ROYVJGckxhZlY4WnBo?=
 =?utf-8?B?QWlURlBzWklwb2c1YzFtcGk3K1ROWUVPZTM2Z01McXd3bTA3Q2hZa0E2OGdQ?=
 =?utf-8?B?WFc4SmhNUXdQTTA4dE42Ylh2K1B6Z3Z0M0NmUFB2MkM0MU0vSjhMSFhWR3Nw?=
 =?utf-8?B?Q1ppYW10WjVyTGdRb3RkMWFHUGdsVkduU0FwcnNUVW5jVG00QnJUS2UwNkNo?=
 =?utf-8?B?NVhmeDVEMFpERmFrUU1mckNRdmtuYVBvWXZadGlHeHNHWFM1QVg5ekI4bG5L?=
 =?utf-8?B?Q3pKNm5TWWw1bGFab3RLVlpkNTZxeGhuMVAzV2Z6MkV1dUdESEpYQWE0ckor?=
 =?utf-8?B?MitkODhQUzdiWEJzZ2dpeEhwa0Rad2FFaDgrYjNnVXRGSEJidWV2NXhpdklv?=
 =?utf-8?B?OHZXMlUvV3d1MEJ1UUltZXBiUll3UkVnbEVma1BnM0I3ZG9pVzBsQzdmYlda?=
 =?utf-8?B?bysvRzhqQlF0R1hTMkxGT3U2T1VudStaZXgzK0FoTGVSaHAzUEl5NHlqaTRI?=
 =?utf-8?B?dDR1Qk83cCtJQzJrTk0wWXRVMVlMWjkyVGRpWFFBUStZN3N4TTBUNDQzT293?=
 =?utf-8?B?UFJaODJkNHZuNDg3Qy9rSisrK2MrenhwUTgxOWc2cFNDVXYvWS9HT3NodHFh?=
 =?utf-8?B?aW9RbEw4RmdmZjlZdHp3RUw3K2laSjBrb2JFZCsreFV0QVJBSDd2Y3BKT2lJ?=
 =?utf-8?B?SzRmMVZSd2ZqVCtIM0VQK2s0RWU4Y1YvcC8yT080NUlRQnZnOUlzWUpCaFpk?=
 =?utf-8?B?YmlXWGJmSXZvdmdqSHY3a2Q1TVdQaTRieTRneElqVzF2QmlHSytMMXRlWnFQ?=
 =?utf-8?B?Mit1RXkzbmdYajRsVjNIb1U0b2hJak9Ud2swT2VGM21EU1lXbDhjcVowdkpy?=
 =?utf-8?B?YlVkNHpNQ0NhMUg0SU45cUdpaWtwVGRpdnJNNUJDMUp6OUpDWTcva09xaGtW?=
 =?utf-8?B?Z2p1UmFGaTR3bGRHRkZPR0FqbExNWFJsOForWFd5V09VRnUvNUQ3cWZwem4y?=
 =?utf-8?B?S0xlMEdDQzVwM2FvRWl6bVhieTYxaEdLWjY3M09nRk9peVErZDJDbEZYbzdp?=
 =?utf-8?B?RUpTb25ubFV2ZHlJZnlEZlh0VFkweTFweEwvUmJvbzRwcFdMdklOdFhpRzFt?=
 =?utf-8?B?QXEyWnZ3VUNTOWNybElaWUZBUzN1c3gwK05FNkhTamFqV1pmeE16M1d2dnpS?=
 =?utf-8?B?WWV2Yithb01QNi9wYnp1aFAwT0hDNnk1MXhyL0ZYQ3FiYXU1bTNzMUNsdkRx?=
 =?utf-8?B?RzNsdUJpWExEZDFKRTBRVnJLZkhBTVJiRmlMQjNXRkVSYytzL2VXaStPMzh4?=
 =?utf-8?B?Nk1kMmI3M1MvTkFKNUlLeGg3eGNMU20wNlBTVE5HcENkTVZjL1g1Wkoza05K?=
 =?utf-8?B?QTBGMEhRZ09iQllRek9wYXovbGozM3dJcFZpdkxMdzN5Zk5MR2dpT3RYKzMy?=
 =?utf-8?B?SlNZNUt3UGdRRVB1SjgxTlorcFZBajJPSHJLTFI2K2srSGo2dlZIaVdFdUZv?=
 =?utf-8?B?T09PN25uSnNzRmhLWC84S1VNaDNoYnVTUEc3VDBUZld2Wnc5MmJJR2RLRncw?=
 =?utf-8?Q?mmjM089eJyZMmDP/3ioiDkI3i?=
X-OriginatorOrg: amd.com
X-MS-Exchange-CrossTenant-Network-Message-Id: a12fe400-5154-48bd-55e9-08dc2d724439
X-MS-Exchange-CrossTenant-AuthSource: BN8PR12MB3108.namprd12.prod.outlook.com
X-MS-Exchange-CrossTenant-AuthAs: Internal
X-MS-Exchange-CrossTenant-OriginalArrivalTime: 14 Feb 2024 15:33:17.4893
 (UTC)
X-MS-Exchange-CrossTenant-FromEntityHeader: Hosted
X-MS-Exchange-CrossTenant-Id: 3dd8961f-e488-4e60-8e11-a82d994e183d
X-MS-Exchange-CrossTenant-MailboxType: HOSTED
X-MS-Exchange-CrossTenant-UserPrincipalName: eDNt+ISZVUyZHaSXO8g9pRNPQGC2MbCeP7SKBaUDQvaH8j2doa+yWZTR3yULAAyf34RBmM+DVCGkJnDhqhO1Ow==
X-MS-Exchange-Transport-CrossTenantHeadersStamped: DM6PR12MB4218

On 2/14/2024 7:02 AM, Borislav Petkov wrote:
> On Tue, Feb 13, 2024 at 09:35:16PM -0600, Yazen Ghannam wrote:
>> +static unsigned int get_cpu_for_fru_num(unsigned int i)
>> +{
>> +	unsigned int cpu = 0;
>> +
>> +	/* Should there be more robust error handling if none found? */
>> +	for_each_online_cpu(cpu) {
> 
> You need to block CPU hotplug operations before iterating over online
> CPUs: cpus_read_lock/unlock.
>

Okay.
  
>> +		if (topology_physical_package_id(cpu) == i)
>> +			return cpu;
>> +	}
>> +
>> +	return cpu;
>> +}
> 
> Fold this one into its single callsite.
>

Ack.
  
>> +
>> +static void init_fmps(void)
>> +{
>> +	struct fru_rec *rec;
>> +	unsigned int i, cpu;
>> +
>> +	for_each_fru(i, rec) {
>> +		cpu = get_cpu_for_fru_num(i);
>> +		set_fmp_fields(get_fmp(rec), cpu);
>> +	}
>> +}
>> +
>> +static int get_system_info(void)
>> +{
>> +	u8 model = boot_cpu_data.x86_model;
> 
> No need for that local var - just use boot_cpu_data.x86_model.
>

Ack.
  
>> +	/* Only load on MI300A systems for now. */
>> +	if (!(model >= 0x90 && model <= 0x9f))
>> +		return -ENODEV;
>> +
>> +	if (!cpu_feature_enabled(X86_FEATURE_AMD_PPIN)) {
>> +		pr_debug("PPIN feature not available");
>> +		return -ENODEV;
>> +	}
>> +
>> +	/* Use CPU Package (Socket) as FRU for MI300 systems. */
>> +	max_nr_fru = topology_max_packages();
>> +	if (!max_nr_fru)
>> +		return -ENODEV;
>> +
>> +	if (!max_nr_entries)
>> +		max_nr_entries = FMPM_DEFAULT_MAX_NR_ENTRIES;
>> +
>> +	max_rec_len  = sizeof(struct fru_rec);
>> +	max_rec_len += sizeof(struct cper_fru_poison_desc) * max_nr_entries;
>> +
>> +	pr_debug("max_nr_fru=%u max_nr_entries=%u, max_rec_len=%lu",
>> +		 max_nr_fru, max_nr_entries, max_rec_len);
>> +	return 0;
>> +}
>> +
>> +static void deallocate_records(void)
> 
> free_records
>

Ack.
  
>> +{
>> +	struct fru_rec *rec;
>> +	int i;
>> +
>> +	for_each_fru(i, rec)
>> +		kfree(rec);
>> +
>> +	kfree(fru_records);
>> +}
>> +
>> +static int allocate_records(void)
>> +{
>> +	int i, ret = 0;
>> +
>> +	fru_records = kcalloc(max_nr_fru, sizeof(struct fru_rec *), GFP_KERNEL);
>> +	if (!fru_records) {
>> +		ret = -ENOMEM;
>> +		goto out;
>> +	}
>> +
>> +	for (i = 0; i < max_nr_fru; i++) {
>> +		fru_records[i] = kzalloc(max_rec_len, GFP_KERNEL);
>> +		if (!fru_records[i]) {
>> +			ret = -ENOMEM;
>> +			goto out_free;
>> +		}
>> +	}
>> +
>> +	return ret;
>> +
>> +out_free:
>> +	for (; i >= 0; i--)
>> +		kfree(fru_records[i]);
>> +
>> +	kfree(fru_records);
>> +out:
>> +	return ret;
>> +}
>> +
>> +static const struct x86_cpu_id fmpm_cpuids[] = {
>> +	X86_MATCH_VENDOR_FAM(AMD, 0x19, NULL),
> 
> This is why this should depend on AMD in Kconfig.
> 

Okay.

I was also thinking that MODULE_DEVICE_TABLE shouldn't be used. Not all
MI300-based systems will need or can use this module. And it does depend
on specific platform configurations.

So the module should not autoload. Users will need to manually load it if
they know that it's usable on their platform. We can keep the cpuid[] and
model checks just for extra safety.

Thanks,
Yazen