From: Peng Zhang <zhangpeng.00@bytedance.com>
To: "Liam R. Howlett" <Liam.Howlett@Oracle.com>
Cc: Peng Zhang <zhangpeng.00@bytedance.com>,
akpm@linux-foundation.org, linux-mm@kvack.org,
linux-kernel@vger.kernel.org, maple-tree@lists.infradead.org
Subject: Re: [RESEND PATCH 3/8] maple_tree: make mas_validate_gaps() to check metadata
Date: Mon, 10 Jul 2023 17:44:47 +0800 [thread overview]
Message-ID: <3c260876-2ee5-7659-0c02-17a68277567f@bytedance.com> (raw)
In-Reply-To: <20230707144550.eavqm7zcr6dofl73@revolver>
在 2023/7/7 22:45, Liam R. Howlett 写道:
> * Peng Zhang <zhangpeng.00@bytedance.com> [230707 06:11]:
>> Make mas_validate_gaps() check whether the offset in the metadata points
>> to the largest gap. By the way, simplify this function.
>>
>> Signed-off-by: Peng Zhang <zhangpeng.00@bytedance.com>
>> ---
>> lib/maple_tree.c | 68 +++++++++++++++++++++++-------------------------
>> 1 file changed, 33 insertions(+), 35 deletions(-)
>>
>> diff --git a/lib/maple_tree.c b/lib/maple_tree.c
>> index 6a8982146338..1fe8b6a787dd 100644
>> --- a/lib/maple_tree.c
>> +++ b/lib/maple_tree.c
>> @@ -6983,15 +6983,16 @@ EXPORT_SYMBOL_GPL(mt_dump);
>> static void mas_validate_gaps(struct ma_state *mas)
>> {
>> struct maple_enode *mte = mas->node;
>> - struct maple_node *p_mn;
>> + struct maple_node *p_mn, *node = mte_to_node(mte);
>> + enum maple_type mt = mte_node_type(mas->node);
>> unsigned long gap = 0, max_gap = 0;
>> unsigned long p_end, p_start = mas->min;
>> - unsigned char p_slot;
>> + unsigned char p_slot, offset;
>> unsigned long *gaps = NULL;
>> - unsigned long *pivots = ma_pivots(mte_to_node(mte), mte_node_type(mte));
>> + unsigned long *pivots = ma_pivots(node, mt);
>> int i;
>>
>> - if (ma_is_dense(mte_node_type(mte))) {
>> + if (ma_is_dense(mt)) {
>> for (i = 0; i < mt_slot_count(mte); i++) {
>> if (mas_get_slot(mas, i)) {
>> if (gap > max_gap)
>> @@ -7004,52 +7005,51 @@ static void mas_validate_gaps(struct ma_state *mas)
>> goto counted;
>> }
>>
>> - gaps = ma_gaps(mte_to_node(mte), mte_node_type(mte));
>> + gaps = ma_gaps(node, mt);
>> for (i = 0; i < mt_slot_count(mte); i++) {
>> - p_end = mas_logical_pivot(mas, pivots, i, mte_node_type(mte));
>> + p_end = mas_logical_pivot(mas, pivots, i, mt);
>>
>> if (!gaps) {
>> - if (mas_get_slot(mas, i)) {
>> - gap = 0;
>> - goto not_empty;
>> - }
>> -
>> - gap += p_end - p_start + 1;
>> + if (!mas_get_slot(mas, i))
>> + gap = p_end - p_start + 1;
>> } else {
>> void *entry = mas_get_slot(mas, i);
>>
>> gap = gaps[i];
>> - if (!entry) {
>> - if (gap != p_end - p_start + 1) {
>> - pr_err("%p[%u] -> %p %lu != %lu - %lu + 1\n",
>> - mas_mn(mas), i,
>> - mas_get_slot(mas, i), gap,
>> - p_end, p_start);
>> - mt_dump(mas->tree, mt_dump_hex);
>> -
>> - MT_BUG_ON(mas->tree,
>> - gap != p_end - p_start + 1);
>> - }
>> - } else {
>> - if (gap > p_end - p_start + 1) {
>> - pr_err("%p[%u] %lu >= %lu - %lu + 1 (%lu)\n",
>> - mas_mn(mas), i, gap, p_end, p_start,
>> - p_end - p_start + 1);
>> - MT_BUG_ON(mas->tree,
>> - gap > p_end - p_start + 1);
>> - }
>> + MT_BUG_ON(mas->tree, !entry);
>> +
>> + if (gap > p_end - p_start + 1) {
>> + pr_err("%p[%u] %lu >= %lu - %lu + 1 (%lu)\n",
>> + mas_mn(mas), i, gap, p_end, p_start,
>> + p_end - p_start + 1);
>> + MT_BUG_ON(mas->tree,
>> + gap > p_end - p_start + 1);
>
> Your change above points out that we are not verifying all gaps are zero
> in non-leaf nodes after p_end >= mas->max. If we don't have a 'no gap'
> indicator then this may be an issue, or maybe it already is an issue?
If we don't have a 'no gap' indicator, why is there an issue? Are you
worried that meta_gap is wrongly pointing to the gap after the node
limit? If so we can verify that meta_gap points to a gap within the node
limit.
>
>> }
>> }
>>
>> if (gap > max_gap)
>> max_gap = gap;
>> -not_empty:
>> +
>> p_start = p_end + 1;
>> if (p_end >= mas->max)
>> break;
>> }
>>
>> counted:
>> + if (mt == maple_arange_64) {
>
> We could loop through the remainder of the gaps here pretty easily.
In this way, it can be verified that the gaps after the node limit are
0.
>
>> + offset = ma_meta_gap(node, mt);
>> + if (offset > mt_slots[mt]) {
>> + pr_err("gap offset %p[%u] is invalid\n", node, offset);
>> + MT_BUG_ON(mas->tree, 1);
>> + }
>> +
>> + if (gaps[offset] != max_gap) {
>> + pr_err("gap %p[%u] is not the largest gap %lu\n",
>> + node, offset, max_gap);
>> + MT_BUG_ON(mas->tree, 1);
>> + }
>> + }
>> +
>> if (mte_is_root(mte))
>> return;
>>
>> @@ -7059,10 +7059,8 @@ static void mas_validate_gaps(struct ma_state *mas)
>> if (ma_gaps(p_mn, mas_parent_type(mas, mte))[p_slot] != max_gap) {
>> pr_err("gap %p[%u] != %lu\n", p_mn, p_slot, max_gap);
>> mt_dump(mas->tree, mt_dump_hex);
>> + MT_BUG_ON(mas->tree, 1);
>> }
>> -
>> - MT_BUG_ON(mas->tree,
>> - ma_gaps(p_mn, mas_parent_type(mas, mte))[p_slot] != max_gap);
>> }
>>
>> static void mas_validate_parent_slot(struct ma_state *mas)
>> --
>> 2.20.1
>>
>>
>> --
>> maple-tree mailing list
>> maple-tree@lists.infradead.org
>> https://lists.infradead.org/mailman/listinfo/maple-tree
next prev parent reply other threads:[~2023-07-10 9:44 UTC|newest]
Thread overview: 24+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-07-07 10:10 [RESEND PATCH 0/8] Improve the validation for maple tree and some cleanup Peng Zhang
2023-07-07 10:10 ` [RESEND PATCH 1/8] maple_tree: set the node limit when creating a new root node Peng Zhang
2023-07-07 15:18 ` Liam R. Howlett
2023-07-10 15:05 ` Liam R. Howlett
2023-07-10 15:07 ` Liam R. Howlett
2023-07-07 10:10 ` [RESEND PATCH 2/8] maple_tree: don't use MAPLE_ARANGE64_META_MAX to indicate no gap Peng Zhang
2023-07-07 15:18 ` Liam R. Howlett
2023-07-07 10:10 ` [RESEND PATCH 3/8] maple_tree: make mas_validate_gaps() to check metadata Peng Zhang
2023-07-07 14:45 ` Liam R. Howlett
2023-07-10 9:44 ` Peng Zhang [this message]
2023-07-10 13:57 ` Liam R. Howlett
2023-07-07 10:10 ` [RESEND PATCH 4/8] maple_tree: fix mas_validate_child_slot() to check last missed slot Peng Zhang
2023-07-07 15:18 ` Liam R. Howlett
2023-07-07 10:10 ` [RESEND PATCH 5/8] maple_tree: make mas_validate_limits() check root node and node limit Peng Zhang
2023-07-07 14:58 ` Liam R. Howlett
2023-07-10 9:11 ` Peng Zhang
2023-07-07 10:10 ` [RESEND PATCH 6/8] maple_tree: update mt_validate() Peng Zhang
2023-07-07 15:02 ` Liam R. Howlett
2023-07-10 9:47 ` Peng Zhang
2023-07-07 10:10 ` [RESEND PATCH 7/8] maple_tree: replace mas_logical_pivot() with mas_safe_pivot() Peng Zhang
2023-07-07 15:19 ` Liam R. Howlett
2023-07-07 10:10 ` [RESEND PATCH 8/8] maple_tree: drop mas_first_entry() Peng Zhang
2023-07-07 15:19 ` Liam R. Howlett
2023-07-07 13:48 ` [RESEND PATCH 0/8] Improve the validation for maple tree and some cleanup Liam R. Howlett
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=3c260876-2ee5-7659-0c02-17a68277567f@bytedance.com \
--to=zhangpeng.00@bytedance.com \
--cc=Liam.Howlett@Oracle.com \
--cc=akpm@linux-foundation.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=maple-tree@lists.infradead.org \
/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