Skip to contentSkip to conversation

boot regression in next-20170428

REGRESSION Niklas Cassel 2 participants 2 messages Apr 28, 2017

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

Hello

Since next-20170428
the ARTPEC-6 SoC (MACH_ARTPEC6) does no longer boot.

It works fine with next-20170427.

I've bisected it to the following commit:

6d684e54690caef45cf14051ddeb7c71beeb681b is the first bad commit commit 6d684e54690caef45cf14051ddeb7c71beeb681b

Author: Herbert Xu <[email protected]>
Date:   Thu Apr 27 13:44:51 2017 +0800

rhashtable: Cap total number of entries to 2^31

When max_size is not set or if it set to a sufficiently large value, the nelems counter can overflow. This would cause havoc with the automatic shrinking as it would then attempt to fit a huge number of entries into a tiny hash table.

This patch fixes this by adding max_elems to struct rhashtable to cap the number of elements. This is set to 2^31 as nelems is not a precise count. This is sufficiently smaller than UINT_MAX that it should be safe.

When max_size is set max_elems will be lowered to at most twice max_size as is the status quo.

Enabling earlyprintk does not give any more info,
it simply hangs after:

JUMP dest=0x2000000 dtb=0x2900000 machine=0x1399 initrd=0x02a00000/0x0021b193 cmdline="earlyprintk debug initcall_debug=1" Uncompressing Linux... done, booting the kernel.

Regards,
Niklas

#1 lore raw

From: Niklas Cassel <[email protected]>
Date: Fri, 28 Apr 2017 17:08:17 +0200

Quoted 9 lines

Hello

Since next-20170428
the ARTPEC-6 SoC (MACH_ARTPEC6) does no longer boot.

It works fine with next-20170427.

Quoted 8 lines

I've bisected it to the following commit:

6d684e54690caef45cf14051ddeb7c71beeb681b is the first bad commit commit 6d684e54690caef45cf14051ddeb7c71beeb681b

Author: Herbert Xu <[email protected]>
Date:   Thu Apr 27 13:44:51 2017 +0800

rhashtable: Cap total number of entries to 2^31

Fixed by:

Quoted 1 line

From 2d2ab658d2debcb4c0e29c9e6f18e5683f3077bf Mon Sep 17 00:00:00 2001

From: Herbert Xu <[email protected]>
Date: Fri, 28 Apr 2017 14:10:48 +0800
Subject: [PATCH] rhashtable: Do not lower max_elems when max_size is zero

The commit 6d684e54690c ("rhashtable: Cap total number of entries to 2^31") breaks rhashtable users that do not set max_size. This is because when max_size is zero max_elems is also incorrectly set to zero instead of 2^31.

This patch fixes it by only lowering max_elems when max_size is not zero.

Fixes 6d684e54690c ("rhashtable: Cap total number of entries to 2^31")
---
 lib/rhashtable.c | 11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)
lib/rhashtable.c +6 -5
index 751630b..3895486 100644
@@ -958,13 +958,14 @@ int rhashtable_init(struct rhashtable *ht,
 	if (params->min_size)
 		ht->p.min_size = roundup_pow_of_two(params->min_size);
 
-	if (params->max_size)
-		ht->p.max_size = rounddown_pow_of_two(params->max_size);
-
 	/* Cap total entries at 2^31 to avoid nelems overflow. */
 	ht->max_elems = 1u << 31;
-	if (ht->p.max_size < ht->max_elems / 2)
-		ht->max_elems = ht->p.max_size * 2;
+
+	if (params->max_size) {
+		ht->p.max_size = rounddown_pow_of_two(params->max_size);
+		if (ht->p.max_size < ht->max_elems / 2)
+			ht->max_elems = ht->p.max_size * 2;
+	}
 
 	ht->p.min_size = max(ht->p.min_size, HASH_MIN_SIZE);
 
Signature · 1 line
2.4.11