linux-kernel - [PATCH v2 1/1] cpufreq: fix the target freq not in the range of policy->min & max

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-next>] [day] [month] [year] [list]

Message-ID: <20210626162324.8236-1-tung-chen.shih@mediatek.com>
Date:   Sun, 27 Jun 2021 00:23:25 +0800
From:   TungChen Shih <tung-chen.shih@...iatek.com>
To:     <rjw@...ysocki.net>, <viresh.kumar@...aro.org>,
        <matthias.bgg@...il.com>
CC:     <wsd_upstream@...iatek.com>, <linux-pm@...r.kernel.org>,
        <linux-kernel@...r.kernel.org>,
        <linux-arm-kernel@...ts.infradead.org>,
        <linux-mediatek@...ts.infradead.org>,
        TungChen Shih <tung-chen.shih@...iatek.com>
Subject: [PATCH v2 1/1] cpufreq: fix the target freq not in the range of policy->min & max

    The function cpufreq_driver_resolve_freq() should return the lowest
supported freq greater than or equal to the given target_freq, subject
to policy (min/max) and driver limitations. However, the index returned
by cpufreq_frequency_table_target() won't subject to policy min/max in
some cases.

    In cpufreq_frequency_table_target(), this function will try to find
an index for @target_freq in freq_table, and the frequency of selected
index should be in the range [policy->min, policy->max], which means:

    policy->min <= policy->freq_table[idx].frequency <= policy->max

    Though "clamp_val(target_freq, policy->min, policy->max);" would
have been called to check this condition, when policy->max or min is
not exactly one of the frequency in the frequency table,
policy->freq_table[idx].frequency may still go out of the range

    For example, if our sorted freq_table is [3000, 2000, 1000], and
suppose we have:

    @target_freq = 2500
    @policy->min = 2000
    @policy->max = 2200
    @relation = CPUFREQ_RELATION_L

1. After clamp_val(target_freq, policy->min, policy->max); @target_freq
becomes 2200
2. Since we use CPUFREQ_REALTION_L, final selected freq will be 3000 which
beyonds policy->max

Signed-off-by: TungChen Shih <tung-chen.shih@...iatek.com>
---
 drivers/cpufreq/cpufreq.c | 15 +++++++++++++++
 1 file changed, 15 insertions(+)

diff --git a/drivers/cpufreq/cpufreq.c b/drivers/cpufreq/cpufreq.c
index 802abc925b2a..8e3a17781618 100644
--- a/drivers/cpufreq/cpufreq.c
+++ b/drivers/cpufreq/cpufreq.c
@@ -544,8 +544,23 @@ unsigned int cpufreq_driver_resolve_freq(struct cpufreq_policy *policy,
 	if (cpufreq_driver->target_index) {
 		unsigned int idx;
 
+		/*  to find the frequency >= target_freq */
 		idx = cpufreq_frequency_table_target(policy, target_freq,
 						     CPUFREQ_RELATION_L);
+
+		/* frequency should subject to policy (min/max) */
+		if (policy->freq_table[idx].frequency > policy->max) {
+			if (policy->freq_table_sorted == CPUFREQ_TABLE_SORTED_ASCENDING)
+				idx--;
+			else
+				idx++;
+		} else if (policy->freq_table[idx].frequency < policy->min) {
+			if (policy->freq_table_sorted == CPUFREQ_TABLE_SORTED_ASCENDING)
+				idx++;
+			else
+				idx--;
+		}
+
 		policy->cached_resolved_idx = idx;
 		return policy->freq_table[idx].frequency;
 	}
-- 
2.18.0