linux-kernel - Re: [PATCH 2/2] RAS: Introduce the FRU Memory Poison Manager

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]

Date: Wed, 14 Feb 2024 19:44:23 +0100
From: Borislav Petkov <bp@...en8.de>
To: Yazen Ghannam <yazen.ghannam@....com>
Cc: tony.luck@...el.com, linux-edac@...r.kernel.org,
	linux-kernel@...r.kernel.org, avadhut.naik@....com,
	john.allen@....com, muralidhara.mk@....com,
	naveenkrishna.chatradhi@....com, sathyapriya.k@....com
Subject: Re: [PATCH 2/2] RAS: Introduce the FRU Memory Poison Manager

On Wed, Feb 14, 2024 at 09:56:14AM -0500, Yazen Ghannam wrote:
> > This one needs to go too.
> > 
> 
> Ack.

Gone:

diff --git a/drivers/ras/amd/fmpm.c b/drivers/ras/amd/fmpm.c
index a67a4b67cf9d..643c36b6dc9c 100644
--- a/drivers/ras/amd/fmpm.c
+++ b/drivers/ras/amd/fmpm.c
@@ -146,11 +146,6 @@ static DEFINE_MUTEX(fmpm_update_mutex);
 #define for_each_fru(i, rec) \
 	for (i = 0; rec = fru_records[i], i < max_nr_fru; i++)
 
-static inline struct cper_fru_poison_desc *get_fpd(struct fru_rec *rec, u32 entry)
-{
-	return &rec->entries[entry];
-}
-
 static inline u32 get_fmp_len(struct fru_rec *rec)
 {
 	return rec->sec_desc.section_length - sizeof(struct cper_section_descriptor);
@@ -253,7 +248,9 @@ static bool rec_has_fpd(struct fru_rec *rec, struct cper_fru_poison_desc *fpd)
 	unsigned int i;
 
 	for (i = 0; i < rec->fmp.nr_entries; i++) {
-		if (same_fpd(get_fpd(rec, i), fpd)) {
+		struct cper_fru_poison_desc *fpd_i = &rec->entries[i];
+
+		if (same_fpd(fpd_i, fpd)) {
 			pr_debug("Found duplicate record");
 			return true;
 		}
@@ -265,7 +262,7 @@ static bool rec_has_fpd(struct fru_rec *rec, struct cper_fru_poison_desc *fpd)
 static void update_fru_record(struct fru_rec *rec, struct mce *m)
 {
 	struct cper_sec_fru_mem_poison *fmp = &rec->fmp;
-	struct cper_fru_poison_desc fpd;
+	struct cper_fru_poison_desc fpd, *fpd_dest;
 	u32 entry = 0;
 
 	mutex_lock(&fmpm_update_mutex);
@@ -287,9 +284,10 @@ static void update_fru_record(struct fru_rec *rec, struct mce *m)
 		goto out_unlock;
 	}
 
-	entry = fmp->nr_entries;
+	entry	  = fmp->nr_entries;
+	fpd_dest  = &rec->entries[entry];
 
-	memcpy(get_fpd(rec, entry), &fpd, sizeof(struct cper_fru_poison_desc));
+	memcpy(fpd_dest, &fpd, sizeof(struct cper_fru_poison_desc));
 
 	fmp->nr_entries		 = entry + 1;
 	fmp->validation_bits	|= FMP_VALID_LIST_ENTRIES;
@@ -359,11 +357,10 @@ static u32 get_cpu_from_fru_id(u64 fru_id)
 
 static void retire_mem_fmp(struct fru_rec *rec, u32 nr_entries, u32 cpu)
 {
-	struct cper_fru_poison_desc *fpd;
 	unsigned int i;
 
 	for (i = 0; i < nr_entries; i++) {
-		fpd = get_fpd(rec, i);
+		struct cper_fru_poison_desc *fpd = &rec->entries[i];
 
 		if (fpd->hw_id_type != FPD_HW_ID_TYPE_MCA_IPID)
 			continue;


> >          /* Use the complement value. */
> >          rec->fmp.checksum = -checksum;
> > 
> > I'd say.
> > 
> 
> This was my first thought. Other checksum code in the kernel does
> the (0-X) thing. So I wasn't sure if there's any odd side effects
> of one over the other. And I didn't take the time to dig into it.

I guess to probably be more expressive? I don't see how

0 - X

and 

-X

differ.

And you can always do a before-after and look at the asm:

before:
# drivers/ras/amd/fmpm.c:202:   rec->fmp.checksum = 0 - checksum;
#NO_APP
        subl    %edx, %eax      # checksum, tmp100
        movl    %eax, 200(%rbx) # tmp100, rec_9(D)->fmp.checksum

after:
# drivers/ras/amd/fmpm.c:202:   rec->fmp.checksum = -checksum;
#NO_APP
        subl    %edx, %eax      # checksum, tmp100
        movl    %eax, 200(%rbx) # tmp100, rec_9(D)->fmp.checksum


> > -/* Calculate a new checksum. */
> > -static u32 get_fmp_checksum(struct fru_rec *rec)
> 
> I made this a helper because we need to validate the checksum when
> reading records from storage too.

It has a single user that's why I whacked it. If a new one materializes,
sure, you can carve it out.

Thx.

-- 
Regards/Gruss,
    Boris.

https://people.kernel.org/tglx/notes-about-netiquette