url: improve port validation#45012
Conversation
|
As with #45011, this could be considered a semver-major breaking change, but we may want to consider this is a bugfix instead. Out of caution, though, we should run CITGM. We should also definitely make sure that |
If a port is not a number, throw rather than treating the `:` that delineates the port as part of the path. This is consistent with WHATWG URL and also mitigates hostname-spoofing. Concerns about hostname-spoofing were raised and presented in excellent detail by pyozzi-toss (pyozzi@toss.im/Security-Tech Team in Toss).
|
CITGM: https://ci.nodejs.org/view/Node.js-citgm/job/citgm-smoker/3019/ (It will 404 until job 3018 finishes, though.) |
|
CITGM results look good to me |
|
Landed in 5f7730e |
If a port is not a number, throw rather than treating the `:` that delineates the port as part of the path. This is consistent with WHATWG URL and also mitigates hostname-spoofing. Concerns about hostname-spoofing were raised and presented in excellent detail by pyozzi-toss (pyozzi@toss.im/Security-Tech Team in Toss). PR-URL: #45012 Reviewed-By: Rafael Gonzaga <rafael.nunu@hotmail.com> Reviewed-By: Yagiz Nizipli <yagiz@nizipli.com> Reviewed-By: Antoine du Hamel <duhamelantoine1995@gmail.com> Reviewed-By: James M Snell <jasnell@gmail.com>
If a port is not a number, throw rather than treating the `:` that delineates the port as part of the path. This is consistent with WHATWG URL and also mitigates hostname-spoofing. Concerns about hostname-spoofing were raised and presented in excellent detail by pyozzi-toss (pyozzi@toss.im/Security-Tech Team in Toss). PR-URL: #45012 Reviewed-By: Rafael Gonzaga <rafael.nunu@hotmail.com> Reviewed-By: Yagiz Nizipli <yagiz@nizipli.com> Reviewed-By: Antoine du Hamel <duhamelantoine1995@gmail.com> Reviewed-By: James M Snell <jasnell@gmail.com>
This ended up breaking |
This reverts commit 5f7730e. This change broke too many edge cases in the ecosystem. Reverting it re-introduces some host-spoofing possibilities, so we won't want to revert forever, but the issue is long-lived enough and not sufficiently critical that we can't wait for a major release to introduce it as a breaking change. After this lands, I plan to re-introduce this as a change that throws a warning rather than an error, after which we can land a semver-major that re-introduces the error and try to get the word out to maintainers of likely-affected packages. Closes: nodejs#45514 Refs: nodejs#45012
|
Given the ecosystem breakage I can't see how this can land in LTS so I've removed the |
This reverts commit 5f7730e. This change broke too many edge cases in the ecosystem. Reverting it re-introduces some host-spoofing possibilities, so we won't want to revert forever, but the issue is long-lived enough and not sufficiently critical that we can't wait for a major release to introduce it as a breaking change. After this lands, I plan to re-introduce this as a change that throws a warning rather than an error, after which we can land a semver-major that re-introduces the error and try to get the word out to maintainers of likely-affected packages. Closes: #45514 Refs: #45012 PR-URL: #45517 Fixes: #45514 Reviewed-By: James M Snell <jasnell@gmail.com> Reviewed-By: Richard Lau <rlau@redhat.com> Reviewed-By: Yagiz Nizipli <yagiz@nizipli.com> Reviewed-By: Antoine du Hamel <duhamelantoine1995@gmail.com>
This reverts commit 5f7730e. This change broke too many edge cases in the ecosystem. Reverting it re-introduces some host-spoofing possibilities, so we won't want to revert forever, but the issue is long-lived enough and not sufficiently critical that we can't wait for a major release to introduce it as a breaking change. After this lands, I plan to re-introduce this as a change that throws a warning rather than an error, after which we can land a semver-major that re-introduces the error and try to get the word out to maintainers of likely-affected packages. Closes: #45514 Refs: #45012 PR-URL: #45517 Fixes: #45514 Reviewed-By: James M Snell <jasnell@gmail.com> Reviewed-By: Richard Lau <rlau@redhat.com> Reviewed-By: Yagiz Nizipli <yagiz@nizipli.com> Reviewed-By: Antoine du Hamel <duhamelantoine1995@gmail.com>
This reverts commit 5f7730e. This change broke too many edge cases in the ecosystem. Reverting it re-introduces some host-spoofing possibilities, so we won't want to revert forever, but the issue is long-lived enough and not sufficiently critical that we can't wait for a major release to introduce it as a breaking change. After this lands, I plan to re-introduce this as a change that throws a warning rather than an error, after which we can land a semver-major that re-introduces the error and try to get the word out to maintainers of likely-affected packages. Closes: nodejs#45514 Refs: nodejs#45012 PR-URL: nodejs#45517 Fixes: nodejs#45514 Reviewed-By: James M Snell <jasnell@gmail.com> Reviewed-By: Richard Lau <rlau@redhat.com> Reviewed-By: Yagiz Nizipli <yagiz@nizipli.com> Reviewed-By: Antoine du Hamel <duhamelantoine1995@gmail.com>
If a port is not a number, throw rather than treating the
:that delineates the port as part of the path. This is consistent with WHATWG URL and also mitigates hostname-spoofing.Concerns about hostname-spoofing were raised and presented in excellent detail by pyozzi-toss (pyozzi@toss.im/Security-Tech Team in Toss).