From 4c9ba9d867020fe37721d1cf11ab6bb0744d2485 Mon Sep 17 00:00:00 2001 From: Andrew Gallant Date: Sun, 14 Mar 2021 14:08:31 -0400 Subject: [PATCH] impl: substantially reduce regex stack size This commit fixes a fairly large regression in the stack size of a Regex introduced in regex 1.4.4. When I dropped thread_local and replaced it with Pool, it turned out that Pool inlined a T into its struct and a Regex in turn had Pool inlined into itself. It further turns out that the T=ProgramCache is itself quite large. We fix this by introducing an indirection in the inner regex type. That is, we use a Box instead of a Pool. This shrinks the size of a Regex from 856 bytes to 16 bytes. Interestingly, prior to regex 1.4.4, a Regex was still quite substantial in size, coming in at around 552 bytes. So it looks like the 1.4.4 release didn't dramatically increase it, but it increased it enough that folks started experiencing real problems: stack overflows. Fixes #750, Fixes #751 Ref https://github.com/servo/servo/pull/28269 --- src/exec.rs | 15 +++++++++++---- 1 file changed, 11 insertions(+), 4 deletions(-) diff --git a/src/exec.rs b/src/exec.rs index 3fd7cc451..3d5a52bea 100644 --- a/src/exec.rs +++ b/src/exec.rs @@ -36,7 +36,14 @@ pub struct Exec { /// All read only state. ro: Arc, /// A pool of reusable values for the various matching engines. - pool: Pool, + /// + /// Note that boxing this value is not strictly necessary, but it is an + /// easy way to ensure that T does not bloat the stack sized used by a pool + /// in the case where T is big. And this turns out to be the case at the + /// time of writing for regex's use of this pool. At the time of writing, + /// the size of a Regex on the stack is 856 bytes. Boxing this value + /// reduces that size to 16 bytes. + pool: Box>, } /// `ExecNoSync` is like `Exec`, except it embeds a reference to a cache. This @@ -1446,11 +1453,11 @@ impl ExecReadOnly { lcs_len >= 3 && lcs_len > self.dfa.prefixes.lcp().char_len() } - fn new_pool(ro: &Arc) -> Pool { + fn new_pool(ro: &Arc) -> Box> { let ro = ro.clone(); - Pool::new(Box::new(move || { + Box::new(Pool::new(Box::new(move || { AssertUnwindSafe(RefCell::new(ProgramCacheInner::new(&ro))) - })) + }))) } }