1048070 - Remove uses of NSPR macros from non-NSS-specific parts of mozilla::pkix

Status: RESOLVED FIXED

(Core :: Security: PSM, defect)

Description

[Brian Smith (:briansmith, :bsmith, use NEEDINFO?)]

Attachment: Part 1: Replace uses of PR_ASSERT with uses of standard assert

Comment 1

[Brian Smith (:briansmith, :bsmith, use NEEDINFO?)]

Attachment: Part 2: Replace uses of PR_NOT_REACHED and PR_ARRAY_SIZE

Note that in this patch, I intentionally didn't change pkixnss.cpp, the NSS-based test code, or the NSS-based name constraint code, because that code uses PR_SetError/SECStatus instead of mozilla::pkix::Result.

Comment 2

[Brian Smith (:briansmith, :bsmith, use NEEDINFO?)]

Attachment: Part 2: Replace uses of PR_NOT_REACHED and PR_ARRAY_SIZE

Comment 3

[Dana Keeler (she/her) [:keeler]]

Comment on attachment 8466842 [details] [diff] [review]
Part 1: Replace uses of PR_ASSERT with uses of standard assert

Review of attachment 8466842 [details] [diff] [review]:
-----------------------------------------------------------------

Looks good to me in general. One question: pkixtestutil.cpp is fairly NSS-heavy - shouldn't we keep using PR_ASSERT there?

::: security/pkix/test/gtest/pkixocsp_VerifyEncodedOCSPResponse.cpp
@@ +391,5 @@
>                const char* signerName,
>                SECOidTag signerEKU = SEC_OID_OCSP_RESPONDER,
>                /*optional, out*/ Input* signerDEROut = nullptr)
>    {
> +    assert(certSubjectName);

Don't we want some sort of gtest macro here?

Comment 4

[Dana Keeler (she/her) [:keeler]]

Comment on attachment 8466845 [details] [diff] [review]
Part 2: Replace uses of PR_NOT_REACHED and PR_ARRAY_SIZE

Review of attachment 8466845 [details] [diff] [review]:
-----------------------------------------------------------------

Looks good.

::: security/pkix/lib/pkixder.h
@@ -36,5 @@
>  // Skip* functions unconditionally advance the input mark and return Success if
>  // they are able to do so; otherwise they fail with the input mark in an
>  // undefined state.
>  
> -#include <cassert>

wait what? Wasn't this added in the previous patch?
Oh, I see - I guess it was moved to Result.h? Sounds good.

Comment 5

[Brian Smith (:briansmith, :bsmith, use NEEDINFO?)]

Comment on attachment 8466842 [details] [diff] [review]
Part 1: Replace uses of PR_ASSERT with uses of standard assert

Review of attachment 8466842 [details] [diff] [review]:
-----------------------------------------------------------------

In pkixtestutil.cpp, I have been removing the NSS dependencies too. My goal is for the test suite to work in the same way as the rest of mozilla::pkix--you plug in a crypto implementation for generating keys and signing and the rest is portable. So, I think it would be better to just keep the pkixtestutil.cpp changes.

::: security/pkix/test/gtest/pkixocsp_VerifyEncodedOCSPResponse.cpp
@@ +391,5 @@
>                const char* signerName,
>                SECOidTag signerEKU = SEC_OID_OCSP_RESPONDER,
>                /*optional, out*/ Input* signerDEROut = nullptr)
>    {
> +    assert(certSubjectName);

If an assertion was OK before, it should be OK now. Though, it is strange that we assert(certSubjectName) but not assert(signerName) even though signerName isn't documented as being /*optional*/. Will make that consistent, at least.

Comment 6

[Brian Smith (:briansmith, :bsmith, use NEEDINFO?)]

Comment on attachment 8466842 [details] [diff] [review]
Part 1: Replace uses of PR_ASSERT with uses of standard assert

Review of attachment 8466842 [details] [diff] [review]:
-----------------------------------------------------------------

::: security/pkix/test/gtest/pkixocsp_VerifyEncodedOCSPResponse.cpp
@@ +391,5 @@
>                const char* signerName,
>                SECOidTag signerEKU = SEC_OID_OCSP_RESPONDER,
>                /*optional, out*/ Input* signerDEROut = nullptr)
>    {
> +    assert(certSubjectName);

I checked this, and signerName is not marked /*optional*/ because the caller is supposed to pass |byKey| (an alias for nullptr) or a non-null char*. A little confusing but marking it /*optional*/ would be more confusing. So, I won't change anything here.

Comment 7

[Brian Smith (:briansmith, :bsmith, use NEEDINFO?)]

https://hg.mozilla.org/integration/mozilla-inbound/rev/54c668e916d3
https://hg.mozilla.org/integration/mozilla-inbound/rev/d7ddaaa69b5a

Comment 8

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/mozilla-central/rev/54c668e916d3
https://hg.mozilla.org/mozilla-central/rev/d7ddaaa69b5a