Tarantool development patches archive
 help / color / mirror / Atom feed
From: Sergey Kaplun via Tarantool-patches <tarantool-patches@dev.tarantool.org>
To: Sergey Ostanevich <sergos@tarantool.org>,
	Maxim Kokryashkin <m.kokryashkin@tarantool.org>
Cc: tarantool-patches@dev.tarantool.org
Subject: [Tarantool-patches] [PATCH luajit] ARM64: Fix pcall() error case.
Date: Fri, 10 Feb 2023 16:01:46 +0300	[thread overview]
Message-ID: <20230210130146.30703-1-skaplun@tarantool.org> (raw)

From: Mike Pall <mike>

Reported by Alex Orlenko.

(cherry picked from commit b4b2dce9fc3ffaaaede39b36d06415311e2aa516)

The `pcall()` assembler preambule modifies `RC` (`x28`) (N args * 8)
during the check of the amount of the given arguments. So, this wrong
value using in the `fff_fallback` routine leading to a crash on the
error throwing, because the Lua stack is filled incorrect and can't be

This patch adds the additional comparison before taking the fallback
branch and modifies `RC` only after this branch.

Sergey Kaplun:
* added the description and the test for the problem

Part of tarantool/tarantool#8069

PR: https://github.com/tarantool/tarantool/pull/8295
Branch: https://github.com/tarantool/luajit/tree/skaplun/lj-762-arm64-pcall-no-arg
* https://github.com/tarantool/tarantool/issues/8069
* https://github.com/LuaJIT/LuaJIT/issues/762

 src/vm_arm64.dasc                                 |  3 ++-
 test/tarantool-tests/lj-762-pcall-no-arg.test.lua | 15 +++++++++++++++
 2 files changed, 17 insertions(+), 1 deletion(-)
 create mode 100644 test/tarantool-tests/lj-762-pcall-no-arg.test.lua

diff --git a/src/vm_arm64.dasc b/src/vm_arm64.dasc
index f517a808..e8b63d33 100644
--- a/src/vm_arm64.dasc
+++ b/src/vm_arm64.dasc
@@ -1168,9 +1168,10 @@ static void build_subroutines(BuildCtx *ctx)
   |//-- Base library: catch errors ----------------------------------------
   |.ffunc pcall
+  |   cmp NARGS8:RC, #8
   |  ldrb TMP0w, GL->hookmask
-  |   subs NARGS8:RC, NARGS8:RC, #8
   |   blo ->fff_fallback
+  |   sub NARGS8:RC, NARGS8:RC, #8
   |    mov RB, BASE
   |    add BASE, BASE, #16
   |  ubfx TMP0w, TMP0w, #HOOK_ACTIVE_SHIFT, #1
diff --git a/test/tarantool-tests/lj-762-pcall-no-arg.test.lua b/test/tarantool-tests/lj-762-pcall-no-arg.test.lua
new file mode 100644
index 00000000..6cbfe707
--- /dev/null
+++ b/test/tarantool-tests/lj-762-pcall-no-arg.test.lua
@@ -0,0 +1,15 @@
+local tap = require('tap')
+-- Test file to check error raising for `pcall()` without
+-- arguments. Regardless that the problem is aarch64-specific,
+-- it is good to test it for all arches.
+-- See also https://github.com/LuaJIT/LuaJIT/issues/762.
+local test = tap.test('lj-762-pcall-no-arg')
+local result, err = pcall(pcall)
+test:ok(not result, 'pcall() without args: bad status')
+test:like(err, 'value expected', 'pcall() without args: error message')
+os.exit(test:check() and 0 or 1)

             reply	other threads:[~2023-02-10 13:05 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-02-10 13:01 Sergey Kaplun via Tarantool-patches [this message]
2023-02-15  1:43 ` Maxim Kokryashkin via Tarantool-patches
2023-02-15  6:57   ` Sergey Kaplun via Tarantool-patches
2023-02-15  8:05     ` Maxim Kokryashkin via Tarantool-patches
2023-02-28 12:43 ` Igor Munkin via Tarantool-patches
2023-03-30 17:38 ` Igor Munkin via Tarantool-patches

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20230210130146.30703-1-skaplun@tarantool.org \
    --to=tarantool-patches@dev.tarantool.org \
    --cc=m.kokryashkin@tarantool.org \
    --cc=sergos@tarantool.org \
    --cc=skaplun@tarantool.org \
    --subject='Re: [Tarantool-patches] [PATCH luajit] ARM64: Fix pcall() error case.' \


* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox