Skip to contentSkip to conversation

x86/head64: use pointer type in sizeof

PATCH In review Alexander Kuleshov 3 participants 3 messages Feb 9, 2016 x86

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 x86/head64: use pointer type in sizeof

We changed loops with memset in the 5e9ebbd87a99 commit (x86/boot:
Micro-optimize reset_early_page_tables()). The base for size of memset was size of pud_p/pmd_p, but the they are actually represented as pointers, although they have the same sizes.

---
based on x86/tip/boot

 arch/x86/kernel/head64.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
arch/x86/kernel/head64.c +2 -2
index 35843ca..7793a17 100644
@@ -75,7 +75,7 @@ again:
 		}
 
 		pud_p = (pudval_t *)early_dynamic_pgts[next_early_pgt++];
-		memset(pud_p, 0, sizeof(pud_p) * PTRS_PER_PUD);
+		memset(pud_p, 0, sizeof(*pud_p) * PTRS_PER_PUD);
 		*pgd_p = (pgdval_t)pud_p - __START_KERNEL_map + phys_base + _KERNPG_TABLE;
 	}
 	pud_p += pud_index(address);
@@ -90,7 +90,7 @@ again:
 		}
 
 		pmd_p = (pmdval_t *)early_dynamic_pgts[next_early_pgt++];
-		memset(pmd_p, 0, sizeof(pmd_p) * PTRS_PER_PMD);
+		memset(pmd_p, 0, sizeof(*pmd_p) * PTRS_PER_PMD);
 		*pud_p = (pudval_t)pmd_p - __START_KERNEL_map + phys_base + _KERNPG_TABLE;
 	}
 	pmd = (physaddr & PMD_MASK) + early_pmd_flags;
Signature · 1 line
2.7.0.25.gfc10eb5
#1 lore raw
Quoted 4 lines · Alexander Kuleshov

We changed loops with memset in the 5e9ebbd87a99 commit (x86/boot:
Micro-optimize reset_early_page_tables()). The base for size of memset was size of pud_p/pmd_p, but the they are actually represented as pointers, although they have the same sizes.

That's rather confused. Also the title is confused:

x86/head64: use pointer type in sizeof

what the fix does it to _not_ use a pointer type.

I fixed the title and the changelog, but you really need to be more careful...

Thanks,

Ingo

#2 lore raw

[tip:x86/boot] x86/boot: Use proper array element type in memset( ) size calculation

In reply to Alexander Kuleshov
x86/boot: Use proper array element type in memset() size calculation
Commit-ID:  a91bbe017552b80e12d712c85549b933a62c6ed4
Gitweb:     http://git.kernel.org/tip/a91bbe017552b80e12d712c85549b933a62c6ed4
Author:     Alexander Kuleshov <[email protected]>
AuthorDate: Tue, 9 Feb 2016 19:44:54 +0600
Committer:  Ingo Molnar <[email protected]>
CommitDate: Tue, 9 Feb 2016 14:55:48 +0100

x86/boot: Use proper array element type in memset() size calculation

I changed open coded zeroing loops to explicit memset()s in the following commit:

5e9ebbd87a99 ("x86/boot: Micro-optimize reset_early_page_tables()")

The base for the size argument of memset was sizeof(pud_p/pmd_p), which are pointers - but the initialized array has pud_t/pmd_t elements.

Luckily the two types had the same size, so this did not result in any runtime misbehavior.

Cc Alexander Popov <[email protected]>
Cc Andrey Ryabinin <[email protected]>
Cc Andy Lutomirski <[email protected]>
Cc Andy Shevchenko <[email protected]>
Cc Linus Torvalds <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Thomas Gleixner <[email protected]>
---
 arch/x86/kernel/head64.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
arch/x86/kernel/head64.c +2 -2
index 35843ca..7793a17 100644
@@ -75,7 +75,7 @@ again:
 		}
 
 		pud_p = (pudval_t *)early_dynamic_pgts[next_early_pgt++];
-		memset(pud_p, 0, sizeof(pud_p) * PTRS_PER_PUD);
+		memset(pud_p, 0, sizeof(*pud_p) * PTRS_PER_PUD);
 		*pgd_p = (pgdval_t)pud_p - __START_KERNEL_map + phys_base + _KERNPG_TABLE;
 	}
 	pud_p += pud_index(address);
@@ -90,7 +90,7 @@ again:
 		}
 
 		pmd_p = (pmdval_t *)early_dynamic_pgts[next_early_pgt++];
-		memset(pmd_p, 0, sizeof(pmd_p) * PTRS_PER_PMD);
+		memset(pmd_p, 0, sizeof(*pmd_p) * PTRS_PER_PMD);
 		*pud_p = (pudval_t)pmd_p - __START_KERNEL_map + phys_base + _KERNPG_TABLE;
 	}
 	pmd = (physaddr & PMD_MASK) + early_pmd_flags;