neovim/test/functional/legacy/memory_usage_spec.lua

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

221 lines
6.4 KiB
Lua
Raw Normal View History

local t = require('test.testutil')
local n = require('test.functional.testnvim')()
local clear = n.clear
local eval = n.eval
local eq = t.eq
local feed_command = n.feed_command
local retry = t.retry
local ok = t.ok
local source = n.source
local poke_eventloop = n.poke_eventloop
local load_adjust = n.load_adjust
local write_file = t.write_file
local is_os = t.is_os
local is_ci = t.is_ci
local is_asan = n.is_asan
clear()
feat(extmark): support proper multiline ranges The removes the previous restriction that nvim_buf_set_extmark() could not be used to highlight arbitrary multi-line regions The problem can be summarized as follows: let's assume an extmark with a hl_group is placed covering the region (5,0) to (50,0) Now, consider what happens if nvim needs to redraw a window covering the lines 20-30. It needs to be able to ask the marktree what extmarks cover this region, even if they don't begin or end here. Therefore the marktree needs to be augmented with the information covers a point, not just what marks begin or end there. To do this, we augment each node with a field "intersect" which is a set the ids of the marks which overlap this node, but only if it is not part of the set of any parent. This ensures the number of nodes that need to be explicitly marked grows only logarithmically with the total number of explicitly nodes (and thus the number of of overlapping marks). Thus we can quickly iterate all marks which overlaps any query position by looking up what leaf node contains that position. Then we only need to consider all "start" marks within that leaf node, and the "intersect" set of that node and all its parents. Now, and the major source of complexity is that the tree restructuring operations (to ensure that each node has T-1 <= size <= 2*T-1) also need to update these sets. If a full inner node is split in two, one of the new parents might start to completely overlap some ranges and its ids will need to be moved from its children's sets to its own set. Similarly, if two undersized nodes gets joined into one, it might no longer completely overlap some ranges, and now the children which do needs to have the have the ids in its set instead. And then there are the pivots! Yes the pivot operations when a child gets moved from one parent to another.
2020-11-22 02:10:37 -07:00
if is_asan() then
pending('ASAN build is difficult to estimate memory usage', function() end)
return
elseif is_os('win') then
if is_ci('github') then
pending(
'Windows runners in Github Actions do not have a stable environment to estimate memory usage',
function() end
)
return
elseif eval("executable('wmic')") == 0 then
pending('missing "wmic" command', function() end)
return
end
elseif eval("executable('ps')") == 0 then
pending('missing "ps" command', function() end)
return
end
local monitor_memory_usage = {
memory_usage = function(self)
local handle
if is_os('win') then
handle = io.popen('wmic process where processid=' .. self.pid .. ' get WorkingSetSize')
else
handle = io.popen('ps -o rss= -p ' .. self.pid)
end
return tonumber(handle:read('*a'):match('%d+'))
end,
op = function(self)
retry(nil, 10000, function()
local val = self.memory_usage(self)
if self.max < val then
self.max = val
end
table.insert(self.hist, val)
ok(#self.hist > 20)
local result = {}
for key, value in ipairs(self.hist) do
if value ~= self.hist[key + 1] then
table.insert(result, value)
end
end
table.remove(self.hist, 1)
self.last = self.hist[#self.hist]
eq(1, #result)
end)
end,
dump = function(self)
return 'max: ' .. self.max .. ', last: ' .. self.last
end,
monitor_memory_usage = function(self, pid)
local obj = {
pid = pid,
max = 0,
last = 0,
hist = {},
}
setmetatable(obj, { __index = self })
obj:op()
return obj
end,
}
setmetatable(monitor_memory_usage, {
__call = function(self, pid)
return monitor_memory_usage.monitor_memory_usage(self, pid)
end,
})
describe('memory usage', function()
local tmpfile = 'X_memory_usage'
after_each(function()
os.remove(tmpfile)
end)
local function check_result(tbl, status, result)
if not status then
print('')
for key, val in pairs(tbl) do
print(key, val:dump())
end
error(result)
end
end
before_each(clear)
--[[
Case: if a local variable captures a:000, funccall object will be free
just after it finishes.
]]
--
it('function capture vargs', function()
local pid = eval('getpid()')
local before = monitor_memory_usage(pid)
write_file(
tmpfile,
[[
func s:f(...)
let x = a:000
endfunc
for _ in range(10000)
call s:f(0)
endfor
]]
)
-- TODO: check_result fails if command() is used here. Why? #16064
feed_command('source ' .. tmpfile)
poke_eventloop()
local after = monitor_memory_usage(pid)
-- Estimate the limit of max usage as 2x initial usage.
-- The lower limit can fluctuate a bit, use 97%.
check_result({ before = before, after = after }, pcall(ok, before.last * 97 / 100 < after.max))
check_result({ before = before, after = after }, pcall(ok, before.last * 2 > after.max))
-- In this case, garbage collecting is not needed.
-- The value might fluctuate a bit, allow for 3% tolerance below and 5% above.
-- Based on various test runs.
local lower = after.last * 97 / 100
local upper = after.last * 105 / 100
check_result({ before = before, after = after }, pcall(ok, lower < after.max))
check_result({ before = before, after = after }, pcall(ok, after.max < upper))
end)
--[[
Case: if a local variable captures l: dict, funccall object will not be
free until garbage collector runs, but after that memory usage doesn't
increase so much even when rerun Xtest.vim since system memory caches.
]]
--
it('function capture lvars', function()
local pid = eval('getpid()')
local before = monitor_memory_usage(pid)
write_file(
tmpfile,
[[
if !exists('s:defined_func')
func s:f()
let x = l:
endfunc
endif
let s:defined_func = 1
for _ in range(10000)
call s:f()
endfor
]]
)
feed_command('source ' .. tmpfile)
poke_eventloop()
local after = monitor_memory_usage(pid)
for _ = 1, 3 do
-- TODO: check_result fails if command() is used here. Why? #16064
feed_command('source ' .. tmpfile)
poke_eventloop()
end
local last = monitor_memory_usage(pid)
-- The usage may be a bit less than the last value, use 80%.
-- Allow for 20% tolerance at the upper limit. That's very permissive, but
-- otherwise the test fails sometimes. On FreeBSD we need to be even much
-- more permissive.
local upper_multiplier = is_os('freebsd') and 19 or 12
local lower = before.last * 8 / 10
local upper = load_adjust((after.max + (after.last - before.last)) * upper_multiplier / 10)
check_result({ before = before, after = after, last = last }, pcall(ok, lower < last.last))
check_result({ before = before, after = after, last = last }, pcall(ok, last.last < upper))
end)
it('releases memory when closing windows when folds exist', function()
if is_os('mac') then
pending('macOS memory compression causes flakiness')
end
local pid = eval('getpid()')
source([[
new
" Insert lines
call nvim_buf_set_lines(0, 0, 0, v:false, repeat([''], 999))
" Create folds
normal! gg
for _ in range(500)
normal! zfjj
endfor
]])
poke_eventloop()
local before = monitor_memory_usage(pid)
source([[
" Split and close window multiple times
for _ in range(1000)
split
close
endfor
]])
poke_eventloop()
local after = monitor_memory_usage(pid)
source('bwipe!')
poke_eventloop()
-- Allow for an increase of 10% in memory usage, which accommodates minor fluctuation,
-- but is small enough that if memory were not released (prior to PR #14884), the test
-- would fail.
local upper = before.last * 1.10
check_result({ before = before, after = after }, pcall(ok, after.last <= upper))
end)
end)