cryptography Remove BER in PKCS#7 and PKCS#12

Starting in 45.0.0, we emit warnings on PKCS#7/PKCS#12 that are parsed by OpenSSL but not by our own DER parser. We are now in the "waiting to see how many people file issues" phase.

Once a sufficient period of time/low issue volume has been established, we will remove the OpenSSL codepaths entirely.

Trackers for filed issues

PKCS#7

[ ] Australian Taxation Office
[ ] Sectigo (Invalid SET ordering)
[ ] Microsoft ActiveDirectory Certificate Services CA (Invalid SET ordering)
[ ] ...

PKCS#12

[ ] Apple Pay Payment Token (https://github.com/pyca/cryptography/issues/13034)
[ ] ...

May 18 '25 15:05 alex

I trigger the BER warning when parsing a PKCS#7 response in my application, which integrates with the Australian Government’s digital identity system (myID), administered by the Australian Taxation Office (ATO). Certificates are issued by the ATO Certificate Authority: http://pki.ato.gov.au/policy/ca.html.

Aug 29 '25 00:08 eidorb

@eidorb thank you for reporting this, is there a PKCS#7 structure that you can share publicly with us, or do they contain PII? Are you able to report this to the Australian Government, or do you need our assistance?

Aug 29 '25 12:08 alex

@alex, see the snippet below (no PII, just UUID).

I don't fully understand the issue to be reported -- can you explain or point to a doc explaining the issue? Is it an issue only for users of the cryptography library going forward, or does it pose a risk to the issuing CA? Are we talking about a MITM exploiting BER to execute arbitrary code during parsing?

import base64

from cryptography import x509
from cryptography.hazmat.primitives.serialization import pkcs7
from pydantic import BaseModel


class CertificateResponse(BaseModel):
    id: int
    p7: str
    p10: str
    credentialToken: str
    links: list[dict[str, str]]

    def decode_certificate_chain(self) -> list[x509.Certificate]:
        """Returns certificate chain contained in `p7` field.

        >>> certificate = CertificateResponse(id=0, p7="MIAGCSqGSIb3DQEHAqCAMIACAQExDTALBglghkgBZQMEAgEwgAYJKoZIhvcNAQcBoIAEAAAAAACggDCCBgIwggTqoAMCAQICFB3hbk57eLEgttua1Svz9rPYOJesMA0GCSqGSIb3DQEBCwUAMH4xCzAJBgNVBAYTAkFVMSMwIQYDVQQKDBpBdXN0cmFsaWFuIFRheGF0aW9uIE9mZmljZTEgMB4GA1UECwwXQ2VydGlmaWNhdGlvbiBBdXRob3JpdHkxKDAmBgNVBAMMH0FUTyBTdWIgQ2VydGlmaWNhdGlvbiBBdXRob3JpdHkwHhcNMjUwMzIwMDcyMTM4WhcNMjcwMzIwMDcyMTM3WjBmMQswCQYDVQQGEwJBVTEXMBUGA1UECgwObXlnb3ZpZC5nb3YuYXUxDzANBgNVBAMMBnBvaSBpZDEtMCsGA1UELhMkNmE0YWJhNDItYjg5Ny00NDc4LWI5ODMtNjc4ZjJiYWQxODIxMIIBIjANBgkqhkiG9w0BAQEFAAOCAQ8AMIIBCgKCAQEAlIA0bxl1altezGE4TqFTwxhcTJuaoDnPxGYBmkFIEkwSyMnk+daJh7eDEYDKRUWgvK74vw6duHFYz9ngACS6vGiJCG+1M3hhyXQyh5B5m0Aoa8KvsY9NhvKLQTKp7x6dzUmRWnw4fVnU6UyaemF2Kd1IHOz8yxC6Nb8qSC7x2Oc1ab099mkjFB/GYmUwa0oOGduD8Jh8oS+OXm5lihKNCbKiRXrmYtNGUINmSrihO/gNZQx/FjRkQQgL9WEnpz/MxhSyhZaLP+AI8K1wJeSB7MjS6ycZ0aqWX6WYnQCKobB9NfkaWy0B0I9ANWN0MSIV1ePHMO1zxmN0cTFnU8FE2QIDAQABo4ICjjCCAoowDAYDVR0TAQH/BAIwADAfBgNVHSMEGDAWgBRe1Ha1ieBZfU4PIcVmBuktVbvuSzBDBggrBgEFBQcBAQQ3MDUwMwYIKwYBBQUHMAKGJ2h0dHA6Ly9wa2kuYXRvLmdvdi5hdS9jcmxzL2F0b3N1YmNhLmNydDCCAZQGA1UdIASCAYswggGHMIIBgwYJKiQBxikBAQcBMIIBdDCCAT4GCCsGAQUFBwICMIIBMB6CASwAVQBzAGUAIAB0AGgAaQBzACAAYwBlAHIAdABpAGYAaQBjAGEAdABlACAAbwBuAGwAeQAgAGYAbwByACAAdABoAGUAIABwAHUAcgBwAG8AcwBlACAAcABlAHIAbQBpAHQAdABlAGQAIABpAG4AIAB0AGgAZQAgAGEAcABwAGwAaQBjAGEAYgBsAGUAIABDAGUAcgB0AGkAZgBpAGMAYQB0AGUAIABQAG8AbABpAGMAeQAuACAATABpAG0AaQB0AGUAZAAgAGwAaQBhAGIAaQBsAGkAdAB5ACAAYQBwAHAAbABpAGUAcwAgAC0AIAByAGUAZgBlAHIAIAB0AG8AIAB0AGgAZQAgAEMAZQByAHQAaQBmAGkAYwBhAHQAZQAgAFAAbwBsAGkAYwB5AC4wMAYIKwYBBQUHAgEWJGh0dHA6Ly9wa2kuYXRvLmdvdi5hdS9wb2xpY3kvY2EuaHRtbDATBgNVHSUEDDAKBggrBgEFBQcDAjA4BgNVHR8EMTAvMC2gK6AphidodHRwOi8vcGtpLmF0by5nb3YuYXUvY3Jscy9hdG9zdWJjYS5jcmwwHQYDVR0OBBYEFCedjwuKBn6gZRpMIwmt01E+7QjJMA4GA1UdDwEB/wQEAwIE8DANBgkqhkiG9w0BAQsFAAOCAQEAFQ3/NnDVQhlYEGZVywprXwKbEusk7OoG/7oDs3633gyiX+xCG2gxKO5XZXtTr2rYxH2cuaQ/DMkXhax/u18HRqoGrlvGnD0qrDJThoEXOGh56APXw9AJlm6zK2nwhX2549kE/UfiV7IhxugpxbXV+O9vdF/h+mcCdGJvdHELkJ3ouOTUsm3sBjYkO818tLh2qQrF2adE8XiYHl/uytyJSrwNpEJqUzmLj8NfeIjt6uAsRN7VZJvJ28lqBfYNeW4tl5i6giRVPaGv5oRRWsiaQDwOfvnWzkfcuzRTyQ8KUcHNNQWq1hrOU/4IH1+gSseOXUtOxjvShC4l5nqLQ6byhTCCBw4wggT2oAMCAQICFF5i28KmWnMoY4zEsZLsSDmdn0sdMA0GCSqGSIb3DQEBCwUAMH8xCzAJBgNVBAYTAkFVMSMwIQYDVQQKDBpBdXN0cmFsaWFuIFRheGF0aW9uIE9mZmljZTEgMB4GA1UECwwXQ2VydGlmaWNhdGlvbiBBdXRob3JpdHkxKTAnBgNVBAMMIEFUTyBSb290IENlcnRpZmljYXRpb24gQXV0aG9yaXR5MB4XDTE5MDQwODAxMzA0MloXDTI5MDQwNTAxMzA0MlowfjELMAkGA1UEBhMCQVUxIzAhBgNVBAoMGkF1c3RyYWxpYW4gVGF4YXRpb24gT2ZmaWNlMSAwHgYDVQQLDBdDZXJ0aWZpY2F0aW9uIEF1dGhvcml0eTEoMCYGA1UEAwwfQVRPIFN1YiBDZXJ0aWZpY2F0aW9uIEF1dGhvcml0eTCCASIwDQYJKoZIhvcNAQEBBQADggEPADCCAQoCggEBAMyYJutAt3p60HPSCC044RBfCKAnpUPzsKK3pXS45lKtyOKAjemff9+3sGFvE87+kLXK89/regDQKmnZmhTmos2PBjbPcG/GuJ2L4CjUqiISDMixQQjxyPZ6L3C5VRboZRjSPhUzPggXzQyaMVU+quOk5CeKfU8qID4FHiUD9I4pQtOpolkmws3o4EsjuGoN1hwIZgIUgT6viz+nzvY9UeLQxt0mhMoQ7GLqI+8Pvy+KxgI1crniTMqtpJ91ZeEMLwVZ9jR8704X7RKM6991hHcu8mGSIUDA7XUcpAHaxj0yGf/I/S/apvh9ZVIBFHta3qGBdta0c3bA7I8Z7aG9y3MCAwEAAaOCAoEwggJ9MBIGA1UdEwEB/wQIMAYBAf8CAQAwHwYDVR0jBBgwFoAU+UoZ71xAOZm3VEo99IxuzmIK8SYwRAYIKwYBBQUHAQEEODA2MDQGCCsGAQUFBzAChihodHRwOi8vcGtpLmF0by5nb3YuYXUvY3Jscy9hdG9yb290Y2EuY3J0MIIBlAYDVR0gBIIBizCCAYcwggGDBgkqJAHGKQEBAQEwggF0MIIBPgYIKwYBBQUHAgIwggEwHoIBLABVAHMAZQAgAHQAaABpAHMAIABjAGUAcgB0AGkAZgBpAGMAYQB0AGUAIABvAG4AbAB5ACAAZgBvAHIAIAB0AGgAZQAgAHAAdQByAHAAbwBzAGUAIABwAGUAcgBtAGkAdAB0AGUAZAAgAGkAbgAgAHQAaABlACAAYQBwAHAAbABpAGMAYQBiAGwAZQAgAEMAZQByAHQAaQBmAGkAYwBhAHQAZQAgAFAAbwBsAGkAYwB5AC4AIABMAGkAbQBpAHQAZQBkACAAbABpAGEAYgBpAGwAaQB0AHkAIABhAHAAcABsAGkAZQBzACAALQAgAHIAZQBmAGUAcgAgAHQAbwAgAHQAaABlACAAQwBlAHIAdABpAGYAaQBjAGEAdABlACAAUABvAGwAaQBjAHkALjAwBggrBgEFBQcCARYkaHR0cDovL3BraS5hdG8uZ292LmF1L3BvbGljeS9jYS5odG1sMDkGA1UdHwQyMDAwLqAsoCqGKGh0dHA6Ly9wa2kuYXRvLmdvdi5hdS9jcmxzL2F0b3Jvb3RjYS5jcmwwHQYDVR0OBBYEFF7UdrWJ4Fl9Tg8hxWYG6S1Vu+5LMA4GA1UdDwEB/wQEAwIBxjANBgkqhkiG9w0BAQsFAAOCAgEAM3uMn7F5LhvQGuAZ9cQOSVmgHVEGLY982tUeYr5utoe4UPo3x2UJy66V5E4RgabTBStYh9JUAroR6N4dCUICYkNgO8Vw9KnYlFIiEiriY+dy4fMyDC7pApDUtyrFKUoSAtuxtw764EMPhCWRTZYTs/zz0ayJyjLmcDCT+4uBzukqKgJnMM8oCVjG7gvUiZfMxK87pphWFYX82jWvK02VtTVZvI32DWodZ2fOVu6DNqq1HCEgC27bLIA05SegY5PdFe6NuCJSbWwAyo31fEy/CozHuQo7tzgTRiNPWII97YUt7O1qu6vl69Da1CgKWrileFFiJKdY/PsmI15mJcoyqj+lDd+18/TDcb4yQRgXHzVpiYLkQI34y7Dn8zJasTOLj77ew7V3RlSOHoNprXLQXKHKzNaBvEInS+IrjLWIEcdEuQVT5RKjxDmIl88hKosY431iaFXFZWS/DAAt5I3JNbdwr7E+AJf6G3GqvuGSC9xq9KsP8oQ8m2xwVK/YCpgBY+H+XJYcAOOLqSNYMGm4q7bActSt6g9HjgHNTxzL8HATzYHQ8thAl/U2ssCTvY52gAonWnetVQlcGcYkaflqBuOZg0HYNvby2YFqKAbVk9CEDPd5lW5VMoHiyQdyyrYnSygi5i40XwdRNDKOfHFzFVsiyHnD4aajXOkZhGjtbeowggfQMIIFuKADAgECAhR9Z3eMg8ktGgVxrRgTg/0tj59eBjANBgkqhkiG9w0BAQsFADB/MQswCQYDVQQGEwJBVTEjMCEGA1UECgwaQXVzdHJhbGlhbiBUYXhhdGlvbiBPZmZpY2UxIDAeBgNVBAsMF0NlcnRpZmljYXRpb24gQXV0aG9yaXR5MSkwJwYDVQQDDCBBVE8gUm9vdCBDZXJ0aWZpY2F0aW9uIEF1dGhvcml0eTAeFw0xOTA0MDgwMDM4NTBaFw0zOTA0MDgwMDM4NTBaMH8xCzAJBgNVBAYTAkFVMSMwIQYDVQQKDBpBdXN0cmFsaWFuIFRheGF0aW9uIE9mZmljZTEgMB4GA1UECwwXQ2VydGlmaWNhdGlvbiBBdXRob3JpdHkxKTAnBgNVBAMMIEFUTyBSb290IENlcnRpZmljYXRpb24gQXV0aG9yaXR5MIICIjANBgkqhkiG9w0BAQEFAAOCAg8AMIICCgKCAgEAtfrVqSRHAxo6EoO+tFaJ5PSVem8jcd7lwFqEZnNFO8e0wNpXPx6/i3mVrWpvyUbSCLldbWd8ph5K2Btxdn+anKDKfKzNJ5ZmPDqXYxIiEoB4HfqrmTJIWr0h3MbtZECQYCVOS1JIl9qsTXOxrrYLLOonFyLMxVUCdq4HWDKVRbkTgszzKseckpZAJgIMFK+iiuzZa5nFOEc3KwmmVSIdwL0zYdAHy7VfP4U0HRjkZnvxnXlaqE8tiNlRswmkiMXdG6SZNTJRmZj5zV08pEkog3iSBNxGivWCP1iI/dhax7wMSXdQY/TrpnqmTttnAw6k6CDtX8k7xBHVzCwc+1mArwu5R36zO1QactYMNxXIVqMNWkNgiwQUQYcKTsAZ/YbdzORbFsngX6pbXaJG3ivemaOuqcjW6hQcT3hWGnLJjtCPQzdTOeBfqu1M2hfM2EWu25arcfPdvHsMjosJvWw8lcstOaOQUv0dbJ4gqsoyNnH0/rcggPtqjpTUXLQc/e2kARIR78AqPJYuf96GOpsOyazNJLdVsE1iiLjoj77wyYmQJPQN/4Dl/PiHxUatJMfNtCH5bn5q49tAHPtgnutcViYMunCw3B4J9R6ol0b008BUmqx2ZdwPsQwLIkeJpieFA26cAwfo+mNbS9PTvy9V+dQ0+IOmclhGiCw2tipGD2sCAwEAAaOCAkIwggI+MA8GA1UdEwEB/wQFMAMBAf8wHwYDVR0jBBgwFoAU+UoZ71xAOZm3VEo99IxuzmIK8SYwRAYIKwYBBQUHAQEEODA2MDQGCCsGAQUFBzAChihodHRwOi8vcGtpLmF0by5nb3YuYXUvY3Jscy9hdG9yb290Y2EuY3J0MIIBkwYDVR0gBIIBijCCAYYwggGCBggqJAHGKQEBATCCAXQwggE+BggrBgEFBQcCAjCCATAeggEsAFUAcwBlACAAdABoAGkAcwAgAGMAZQByAHQAaQBmAGkAYwBhAHQAZQAgAG8AbgBsAHkAIABmAG8AcgAgAHQAaABlACAAcAB1AHIAcABvAHMAZQAgAHAAZQByAG0AaQB0AHQAZQBkACAAaQBuACAAdABoAGUAIABhAHAAcABsAGkAYwBhAGIAbABlACAAQwBlAHIAdABpAGYAaQBjAGEAdABlACAAUABvAGwAaQBjAHkALgAgAEwAaQBtAGkAdABlAGQAIABsAGkAYQBiAGkAbABpAHQAeQAgAGEAcABwAGwAaQBlAHMAIAAtACAAcgBlAGYAZQByACAAdABvACAAdABoAGUAIABDAGUAcgB0AGkAZgBpAGMAYQB0AGUAIABQAG8AbABpAGMAeQAuMDAGCCsGAQUFBwIBFiRodHRwOi8vcGtpLmF0by5nb3YuYXUvcG9saWN5L2NhLmh0bWwwHQYDVR0OBBYEFPlKGe9cQDmZt1RKPfSMbs5iCvEmMA4GA1UdDwEB/wQEAwIBxjANBgkqhkiG9w0BAQsFAAOCAgEAGxf6n1gvy9l+fijdkcE2N4pZpfAmc7CQ2v4h6vhqUoqcr/EzPe2K5Sy8BkWkXn+Oc9HAR4SmAzyVi3/5tKq+rBEQ48a7aFqbl7mX1L99xrVrF4A21jAbz0TU0BWJGw5azsPJvReobOe6CGHZFUsyT89TLwxNOObp/E4LkPdkViDAjt54a309NqW6/V1yqfGBn6kTArTHe+eOr45y0MH4AuiVDW1FIL6ZqejMJGICwrABsdxyl6d3VHsA95vRunwVY/k/v6l1eClAAXgV728h0klp8ssjBw1my7NmbxXj3AdPzPHv8knfmxLY4bwmoMKEMRkaprSQtkl99jhq6PJJi2+IB0GoW2Cup8kMLVR5y0zpsThSEumGqngQIJUQtna0uYTc4SAMfhSfOVUyerjqLo/4Ka0FbuTo4frx7k10IEjjvynBrTIOgpKMCYG+3rJEdto7ZPZOCO8cvEzWOOTrK6W7nMs5Ze1l89tgwtO07Sd494/KZWd4qqMIxc2KTAKMqftWPwK8nziivYILYNfRrdj4xwL2I7nlTUuRcMZ0winBBhZz8M8CqO9olVdDzKBICyC0eydADd1UTbnfuVxChMAwJqtOewPQSTfJIQxxemEkAg9Qh2IKiK8g8Fp8uOHnnb8lvdCxMoakwSpvHJ1vHU9MQ6a96rfW4xtzOuO/6vkAADGCAlowggJWAgEBMIGXMH8xCzAJBgNVBAYTAkFVMSMwIQYDVQQKDBpBdXN0cmFsaWFuIFRheGF0aW9uIE9mZmljZTEgMB4GA1UECwwXQ2VydGlmaWNhdGlvbiBBdXRob3JpdHkxKTAnBgNVBAMMIEFUTyBSb290IENlcnRpZmljYXRpb24gQXV0aG9yaXR5AhReYtvCplpzKGOMxLGS7Eg5nZ9LHTALBglghkgBZQMEAgGggZYwGAYJKoZIhvcNAQkDMQsGCSqGSIb3DQEHATAcBgkqhkiG9w0BCQUxDxcNMjUwMzIwMDczMTM4WjArBgkqhkiG9w0BCTQxHjAcMAsGCWCGSAFlAwQCAaENBgkqhkiG9w0BAQsFADAvBgkqhkiG9w0BCQQxIgQg47DEQpj8HBSa+/TImW+5JCeuQeRkm5NMpJWZG3hSuFUwDQYJKoZIhvcNAQELBQAEggEAI0EE0TkdnbTPI30mXRBBJyt46AA7lWhBIDwDdkQJLTF/ZMzSPaTSZ5q2AQiPz0SrH2AClD7gxVeMi6mo/h2gIx8hO0IGkK1+9ul4kTK0MqCRHMi53LljqqEvD4ExBKewDa1/bHiL15KbmwqnxPJ3cu/Vsc+jtVhWeCthZc/k+zIFlcYoHOF3/fIFD+VRs50ohi2G0P0mE3zN4VOBmIT0SERDyPYod4mVIkS7D4fwy7Q2sPINhmg++nG0/PwYXBbD+0fni4ZQ3TVyclGmky5Mg+o9yUyfhkVHkAaD58YA5hgeXkp75pT5H8mlmmAlO2x7OYM7iq4DknT2zEnYzArP1gAAAAAAAA==", p10="...", credentialToken="...", links=[{"...": "..."}])
        >>> certificates = certificate.decode_certificate_chain()
        >>> certificates
        [<Certificate(subject=<Name(C=AU,O=mygovid.gov.au,CN=poi id,2.5.4.46=6a4aba42-b897-4478-b983-678f2bad1821)>, ...)>, <Certificate(subject=<Name(C=AU,O=Australian Taxation Office,OU=Certification Authority,CN=ATO Sub Certification Authority)>, ...)>, <Certificate(subject=<Name(C=AU,O=Australian Taxation Office,OU=Certification Authority,CN=ATO Root Certification Authority)>, ...)>]
        """
        return pkcs7.load_der_pkcs7_certificates(base64.standard_b64decode(self.p7))

Aug 29 '25 21:08 eidorb

The concern here is that it requires us to retain a BER parser. Basically every modern cryptographic protocol uses DER (which is a subset of BER and is more rigid).

In our case, we have a Rust (memory safe) DER parser, and rely on OpenSSL (memory unsafe) for BER parsing. We'd like to migrate PKCS#7 to our Rust DER parser, but can't do that as long as we need to support BER. There's no particular risk to the issuer here, but we think most folks will want to do the right thing and just use DER -- there's no real reason not to.

Aug 29 '25 21:08 alex

Looks like in this case the issue is that they use indefinite length encodings.

Aug 29 '25 21:08 alex

Understood. I will attempt to communicate with the abyss as a courtesy, explaining the above. It may be a simple checkbox or flag in their system's configuration. As mentioned in the other issue, DER is a strict subset of BER, so there should not be compatibility issues for existing consumers. It just makes sense.

Aug 29 '25 21:08 eidorb

Thank you! I'll also see if I can find a contact at the ATO.

On Fri, Aug 29, 2025 at 5:51 PM eidorb @.***> wrote:

eidorb left a comment (pyca/cryptography#12936) https://github.com/pyca/cryptography/issues/12936#issuecomment-3238365296

Understood. I will attempt to communicate with the abyss as a courtesy, explaining the above. It may be a simple checkbox or flag in their system's configuration. As mentioned in the other issue, DER is a strict subset of BER, so there should not be compatibility issues for existing consumers. I just makes sense.

— Reply to this email directly, view it on GitHub https://github.com/pyca/cryptography/issues/12936#issuecomment-3238365296, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAAAGBFSHJBHUCEFX7MTIML3QDDMNAVCNFSM6AAAAAB5L42TQKVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTEMZYGM3DKMRZGY . You are receiving this because you were mentioned.Message ID: @.***>

-- All that is necessary for evil to succeed is for good people to do nothing.

Aug 29 '25 21:08 alex

@vteague do you have a contact who would be across ATO's PKI/myID infra?

myID's list of third party licenses indicate probable use of Dot Net cryptography libraries. It should certainly support DER, and as mentioned above, compatibility issues should be minimal due to DER being a more rigid subset of BER.

(PS: Apologies for tagging you without any context whatsoever 🙇)

Sep 08 '25 02:09 eidorb

+1 for the Apple Pay use case.

> pkcs7.load_der_pkcs7_certificates(data)
UserWarning: PKCS#7 certificates could not be parsed as DER, falling back to parsing as BER. Please file an issue at https://github.com/pyca/cryptography/issues explaining how your PKCS#7 certificates were created. In the future, this may become an exception.

Oct 06 '25 13:10 gryevns

I also had the case with a PKCS#7 S-MIME certificate generated today from Sectigo.

Here is the log:

[07/Oct/2025 11:19:54: WARNING/ForkPoolWorker-1] /app/main/certificates_helpers.py:64: UserWarning: PKCS#7 certificates could not be parsed as DER, falling back to parsing as BER. Please file an issue at https://github.com/pyca/cryptography/issues explaining how your PKCS#7 certificates were created. In the future, this may become an exception.
certs = pkcs7.load_pem_pkcs7_certificates(pkcs7_cert.encode("utf-8"))

Oct 07 '25 09:10 joachimBurket

Are you able to share the PKCS#7 module so we can clearly report the issue to sectigo?

Oct 07 '25 09:10 alex

I'm not sure how to get the module from my PKCS#7 file? (I have access to openssl CLI)

Oct 07 '25 10:10 joachimBurket

I'm asking if you can upload the PKCS#7 file that produces this warning.

Oct 07 '25 10:10 alex

Ah okay sorry I thought it was only a part of the certificate.

Here is a dummy one generated just now:

Oct 07 '25 10:10 joachimBurket

Thanks -- the issue here is that certificates is supposed to be a SET but the values here are not in lexicographic order. I'll report this to Sectigo. Thank you for the detailed report.

Oct 07 '25 11:10 alex

Rob from Sectigo here. Thanks for the report, @alex and @joachimBurket. I'll see what I can do to fix our PKCS#7 DER-encoder.

Oct 09 '25 10:10 robstradling

Thanks @robstradling! We appreciate it!

Oct 09 '25 10:10 reaperhulk

@eidorb I'm sorry for not noticing this thread earlier - no unfortunately I don't have any contacts in the ATO. Did you find anyone? Reporting problems to AusGov is always a challenge, but usually if you can guess your way through the options here https://www.cyber.gov.au/report-and-recover/report there's a possibility it will be read by someone who can pass on the information to where it is needed.

Oct 09 '25 19:10 vteague

Hi All,

I'm also receiving the UserWarning when attempting to parse a PKCS#7 payload returned by a Microsoft ActiveDirectory Certificate Services CA (Windows Server 2022 Standard - build 21H2). The CA also erroneously returns the PKCS#7 with a -----BEGIN CERTIFICATE---- pem block (bum bum bah bum....), but thats for another story. After string replacing the BEGIN/END CERTIFICATE with the appropriate BEGIN/END PKCS7 the UserWarning is generated.

Here's the PKCS#7 data:

Thank you in advance for any (if any) insight!

edit: I forgot to include this is encountered when requesting the "chain of authority" endpoint: https://hostname.domain.tld/certsrv/certnew.p7b?ReqID=CACert&Renewal=1&Enc=b64

Oct 15 '25 15:10 giantg

Looks like this is the same issue as with the Sectigo certs -- the SET is not correctly ordered.

Oct 15 '25 15:10 alex

@alex, this is probably not the place/forum for it, but is there a guide or a way you were able to tell that so quickly? I've used openssl pkcs7 -noout -print to view what appears to be the structure/contents of the PKCS#7 data, but obviously I didn't catch what you did.

Oct 15 '25 15:10 giantg

I hacked up cryptography to show me the specific error instead of just a generic warning -- I'll try to land something that includes this in the warning.

Oct 15 '25 15:10 alex

Unfortunately, GeoTrust (by DigiCert) certificates throw this warning as well.

How do I validate if it's due to the ordering of SET?

Oct 20 '25 10:10 mrestorff

cryptography main (since https://github.com/pyca/cryptography/commit/f5802fe7e08774cb8f8d80ede396de91efb1afca) now includes more details in the warning -- if you're able to build on main it should provide information. Or if you can share the PKCS#7 bundle I can verify.

Oct 21 '25 03:10 alex

Perfect, thanks for the advice. As expected:

Error details: ValueError: error parsing asn1 value: ParseError { kind: InvalidSetOrdering,
location: ["ContentInfo::content", "SignedData::certificates", 1] }

Oct 21 '25 07:10 mrestorff

Awesome -- if you're able to report that to DigiCert it'd be great!

On Tue, Oct 21, 2025 at 12:06 AM mrestorff @.***> wrote:

mrestorff left a comment (pyca/cryptography#12936) https://github.com/pyca/cryptography/issues/12936#issuecomment-3425057986

Perfect, thanks for the advice. As expected:

Error details: ValueError: error parsing asn1 value: ParseError { kind: InvalidSetOrdering, location: ["ContentInfo::content", "SignedData::certificates", 1] }

— Reply to this email directly, view it on GitHub https://github.com/pyca/cryptography/issues/12936#issuecomment-3425057986, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAAAGBHQ6JQBRKIC4T3GRCL3YXLODAVCNFSM6AAAAAB5L42TQKVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTIMRVGA2TOOJYGY . You are receiving this because you were mentioned.Message ID: @.***>

-- All that is necessary for evil to succeed is for good people to do nothing.

Oct 21 '25 13:10 alex

I've encountered this with S/MIME (smime.p7s) generated by Outlook (Classic). The .p7s contains "DigiCert Assured ID Root G2", "DigiCert Assured G2 SMIME RSA4096 SHA384 2024 CA1" and one personal certificate under that CA.

Using latest master (https://github.com/pyca/cryptography/commit/50c9a7c2428c8b2f63b027616790895fbb673c39) it gives the following error:

 ValueError: error parsing asn1 value: ParseError { kind: InvalidLength }

The fallback works.

Dec 09 '25 21:12 TaaviE

Looking into this more it seems like Outlook S/MIME signing outputs a PKCS#7 structure with indefinite length EncapsulatedContentInfo even if no eContent is used in the SEQUENCE.

The CMS RFC says the following:

Signed attributes and authenticated attributes are the only data types used in the CMS that require DER encoding.

The section about EncapsulatedContentInfo says:

The content is represented in the type EncapsulatedContentInfo eContentType is an object identifier. The object identifier uniquely specifies the content type. eContent is the content itself, carried as an octet string. The eContent need not be DER encoded.

While it is somewhat lazy that Outlook doesn't use definite lengths on .p7s S/MIME signatures (with no eContent), it seems to be correct based on the CMS RFC. Which means that if cryptography removes BER fallback it will likely make S/MIME PKCS#7 very annoying to handle. Even if it's for example just to read the certificates within.

Having a memory-safe fallback BER parser would keep this use-case functional.

Dec 12 '25 14:12 TaaviE