This should be a relatively straightforward patch. Essentially, we implement
a wrapper class (and friends) around nsZipArchive (and friends), which transparently
caches entries from the underlying zip archive in the StartupCache. This will break
without changes to the StartupCache, made in the patch after this, which allow it
to be used off of the main thread, and outside the main process.
Depends on D77635
Differential Revision: https://phabricator.services.mozilla.com/D77634
The overall goal of this patch is to make the StartupCache accessible anywhere.
There's two main pieces to that equation:
1. Allowing it to be accessed off main thread, which means modifying the
mutex usage to ensure that all data accessed from non-main threads is
protected.
2. Allowing it to be accessed out of the chrome process, which means passing
a handle to a shared cache buffer down to child processes.
Number 1 is somewhat fiddly, but it's all generally straightforward work. I'll
hope that the comments and the code are sufficient to explain what's going on
there.
Number 2 has some decisions to be made:
- The first decision was to pass a handle to a frozen chunk of memory down to
all child processes, rather than passing a handle to an actual file. There's
two reasons for this: 1) since we want to compress the underlying file on
disk, giving that file to child processes would mean they have to decompress
it themselves, eating CPU time. 2) since they would have to decompress it
themselves, they would have to allocate the memory for the decompressed
buffers, meaning they cannot all simply share one big decompressed buffer.
- The drawback of this decision is that we have to load and decompress the
buffer up front, before we spawn any child processes. We attempt to
mitigate this by keeping track of all the entries that child processes
access, and only including those in the frozen decompressed shared buffer.
- We base our implementation of this approach off of the shared preferences
implementation. Hopefully I got all of the pieces to fit together
correctly. They seem to work in local testing and on try, but I think
they require a set of experienced eyes looking carefully at them.
- Another decision was whether to send the handles to the buffers over IPC or
via command line. We went with the command line approach, because the startup
cache would need to be accessed very early on in order to ensure we do not
read from any omnijars, and we could not make that work via IPC.
- Unfortunately this means adding another hard-coded FD, similar to
kPrefMapFileDescriptor. It seems like at the very least we need to rope all
of these together into one place, but I think that should be filed as a
follow-up?
Lastly, because this patch is a bit of a monster to review - first, thank you
for looking at it, and second, the reason we're invested in this is because we
saw a >10% improvement in cold startup times on reference hardware, with a p
value less than 0.01. It's still not abundantly clear how reference hardware
numbers translate to numbers on release, and they certainly don't translate
well to Nightly numbers, but it's enough to convince me that it's worth some
effort.
Depends on D78584
Differential Revision: https://phabricator.services.mozilla.com/D77635
Opening our Omnijars can be expensive, and it should be deferrable until after
startup is completed, provided we have a startup cache. In a previous patch in this
stack, we implemented caching of the zip central directory for omnijars, but we
still have to open the file in order to hand the object off to various omnijar
consumers. In a later patch, we will wrap nsZipArchive access in a class which
will allow us to transparently cache nsZipArchive results. These two get us
most of the way to not needing to read from the underlying omnijar files during
startup, but there are still nontrivial pieces, like nsZipFind for instance,
which we don't want to just duplicate inside of a wrapper class, so we would
like to sort out a way in which we can use an nsZipArchive class, but not
actually back it up with the real underlying file until we really need data
from it which we can't find in a cache.
Depends on D77633
Differential Revision: https://phabricator.services.mozilla.com/D78584
We need to be able to init StartupCache before the Omnijar in order to cache
all of the Omnijar contents we access. This patch implements that.
Depends on D77632
Differential Revision: https://phabricator.services.mozilla.com/D77633
We would like to be able to defer opening the omnijar files until after startup
if the StartupCache has already been populated. Opening the omnijar files takes
a nontrivial time, at least on Windows, and almost everything in the omnijar
should be fairly compressible, and thus makes sense to live in the StartupCache.
See the last patch in this series for a little more discussion on numbers, but
tl;dr: we saw a 12% improvement in time to about:home being finished on reference
hardware with these changes together with the changes from the descendant patches.
Differential Revision: https://phabricator.services.mozilla.com/D77632
NS_INLINE_DECL_REFCOUNTING macro doesn't properly work when the object is used on a thread that isn't backed by a single PRThread (such as TaskQueue). See bug 1648031.
The resolution of this issue is rather complex, and outside the scope of this series of change.
So for now, we create a new macro NS_INLINE_DECL_REFCOUNTING_ONEVENTTHREAD which will use a different mechanism to ensure the thread-safe usage of a class.
Differential Revision: https://phabricator.services.mozilla.com/D81269
This should be a relatively straightforward patch. Essentially, we implement
a wrapper class (and friends) around nsZipArchive (and friends), which transparently
caches entries from the underlying zip archive in the StartupCache. This will break
without changes to the StartupCache, made in the patch after this, which allow it
to be used off of the main thread, and outside the main process.
Depends on D77635
Differential Revision: https://phabricator.services.mozilla.com/D77634
The overall goal of this patch is to make the StartupCache accessible anywhere.
There's two main pieces to that equation:
1. Allowing it to be accessed off main thread, which means modifying the
mutex usage to ensure that all data accessed from non-main threads is
protected.
2. Allowing it to be accessed out of the chrome process, which means passing
a handle to a shared cache buffer down to child processes.
Number 1 is somewhat fiddly, but it's all generally straightforward work. I'll
hope that the comments and the code are sufficient to explain what's going on
there.
Number 2 has some decisions to be made:
- The first decision was to pass a handle to a frozen chunk of memory down to
all child processes, rather than passing a handle to an actual file. There's
two reasons for this: 1) since we want to compress the underlying file on
disk, giving that file to child processes would mean they have to decompress
it themselves, eating CPU time. 2) since they would have to decompress it
themselves, they would have to allocate the memory for the decompressed
buffers, meaning they cannot all simply share one big decompressed buffer.
- The drawback of this decision is that we have to load and decompress the
buffer up front, before we spawn any child processes. We attempt to
mitigate this by keeping track of all the entries that child processes
access, and only including those in the frozen decompressed shared buffer.
- We base our implementation of this approach off of the shared preferences
implementation. Hopefully I got all of the pieces to fit together
correctly. They seem to work in local testing and on try, but I think
they require a set of experienced eyes looking carefully at them.
- Another decision was whether to send the handles to the buffers over IPC or
via command line. We went with the command line approach, because the startup
cache would need to be accessed very early on in order to ensure we do not
read from any omnijars, and we could not make that work via IPC.
- Unfortunately this means adding another hard-coded FD, similar to
kPrefMapFileDescriptor. It seems like at the very least we need to rope all
of these together into one place, but I think that should be filed as a
follow-up?
Lastly, because this patch is a bit of a monster to review - first, thank you
for looking at it, and second, the reason we're invested in this is because we
saw a >10% improvement in cold startup times on reference hardware, with a p
value less than 0.01. It's still not abundantly clear how reference hardware
numbers translate to numbers on release, and they certainly don't translate
well to Nightly numbers, but it's enough to convince me that it's worth some
effort.
Depends on D78584
Differential Revision: https://phabricator.services.mozilla.com/D77635
Opening our Omnijars can be expensive, and it should be deferrable until after
startup is completed, provided we have a startup cache. In a previous patch in this
stack, we implemented caching of the zip central directory for omnijars, but we
still have to open the file in order to hand the object off to various omnijar
consumers. In a later patch, we will wrap nsZipArchive access in a class which
will allow us to transparently cache nsZipArchive results. These two get us
most of the way to not needing to read from the underlying omnijar files during
startup, but there are still nontrivial pieces, like nsZipFind for instance,
which we don't want to just duplicate inside of a wrapper class, so we would
like to sort out a way in which we can use an nsZipArchive class, but not
actually back it up with the real underlying file until we really need data
from it which we can't find in a cache.
Depends on D77633
Differential Revision: https://phabricator.services.mozilla.com/D78584
We need to be able to init StartupCache before the Omnijar in order to cache
all of the Omnijar contents we access. This patch implements that.
Depends on D77632
Differential Revision: https://phabricator.services.mozilla.com/D77633
We would like to be able to defer opening the omnijar files until after startup
if the StartupCache has already been populated. Opening the omnijar files takes
a nontrivial time, at least on Windows, and almost everything in the omnijar
should be fairly compressible, and thus makes sense to live in the StartupCache.
See the last patch in this series for a little more discussion on numbers, but
tl;dr: we saw a 12% improvement in time to about:home being finished on reference
hardware with these changes together with the changes from the descendant patches.
Differential Revision: https://phabricator.services.mozilla.com/D77632
We want it to returning the actual nsThread if that's where the MessageLoop would dispatch its tasks; otherwise return the MessageLoop's EventTarget
Depends on D80357
Differential Revision: https://phabricator.services.mozilla.com/D80811
NS_INLINE_DECL_REFCOUNTING macro doesn't properly work when the object is used on a thread that isn't backed by a single PRThread (such as TaskQueue). See bug 1648031.
The resolution of this issue is rather complex, and outside the scope of this series of change.
So for now, we create a new macro NS_INLINE_DECL_REFCOUNTING_ONEVENTTHREAD which will use a different mechanism to ensure the thread-safe usage of a class.
Differential Revision: https://phabricator.services.mozilla.com/D81269
We want it to returning the actual nsThread if that's where the MessageLoop would dispatch its tasks; otherwise return the MessageLoop's EventTarget
Depends on D80357
Differential Revision: https://phabricator.services.mozilla.com/D80811
Apparently I added these in the initial commit for RecursiveMutex. I'm
not quite sure what I was thinking, but we don't need them for the
RecursiveMutex itself. (We have them on the corresponding `*Auto*Lock`
classes, which are also `MOZ_RAII`.)
Differential Revision: https://phabricator.services.mozilla.com/D81345
This just spins the event loop during fast shutdown until all queued
cert_storage tasks have completed. The patch achieves this by simply
adding a counter which will be incremented and decremented on the
main thread via tying into the tasks' `new` and `done` methods. A
slightly more performant solution would use a condvar and sleep the
main thread waiting on pending operations to complete, but given the
low frequency of these occuring during shutdown, such an approach
would be overkill.
Differential Revision: https://phabricator.services.mozilla.com/D80906
We were storing LogModule::mName as an &'static str in the map... That's
not fine.
The Arc shenanigans were also more complicated than they need to be IMO.
This uses a plain atomic bool to keep the fast path snappy.
Differential Revision: https://phabricator.services.mozilla.com/D81007
