> Note that ContentParent runs in the parent/chrome process. This is probably not what you want? Oh you're right... that definitely seems like the wrong place. Would you have any thoughts on the appropriate location in the code to do this preloading? I'm hoping to find the place in source which runs in the content process prior to the content sandbox enforcement. > There are some differences between running a normal (packaged) Firefox and running the build output. This is good to know! For the moment, we are still in the phase where we are testing on build output only. I will look definitely into this when we start testing the packaged builds. > I'm not sure if this is different for mach run vs mach test. Looking at MOZ_SANDBOX_LOGGING output will tell you if the expected paths were whitelisted. Thanks, I will try this out. > typically have their return values ignored in the content process Unfortunately the call to canonicalize is in third party rust code, which does abort on failure... From your description. it seems possible that this call to canonicalize may be one of the sources of problems for `mach test`... I will investigate and check if this needs to be removed.
Bug 1575985 Comment 12 Edit History
Note: The actual edited comment in the bug view page will always show the original commenter’s name and original timestamp.
> Note that ContentParent runs in the parent/chrome process. This is probably not what you want? Oh you're right... that definitely seems like the wrong place. Would you have any thoughts on the appropriate location in the code to do this preloading? I'm hoping to find the place in source which runs in the content process prior to the content sandbox enforcement. > There are some differences between running a normal (packaged) Firefox and running the build output. This is good to know! For the moment, we are still in the phase where we are testing on build output only. I will look definitely into this when we start testing the packaged builds. > I'm not sure if this is different for mach run vs mach test. Looking at MOZ_SANDBOX_LOGGING output will tell you if the expected paths were whitelisted. Thanks, I will try this out. > typically have their return values ignored in the content process Unfortunately the call to canonicalize is in third party rust code, which does abort on failure... From your description, it seems possible that this call to canonicalize may be one of the sources of problems for `mach test`... I will investigate and check if this needs to be removed.