sfx2/source/doc/objmisc.cxx | 10 +++++++++-
1 file changed, 9 insertions(+), 1 deletion(-)
New commits:
commit 71aa5352f6f57283d1db51c980f6f0b802381871
Author: Miklos Vajna <vmik...@collabora.com>
AuthorDate: Thu Jan 23 12:58:10 2025 +0100
Commit: Miklos Vajna <vmik...@collabora.com>
CommitDate: Thu Jan 23 19:39:23 2025 +0100
sfx2: fix uncaught exception in SfxObjectShell::isEditDocLocked()
Crashreport signature:
SIG Fatal signal received: SIGABRT code: 18446744073709551610 for
address: 0x1da00001e2c
program/libmergedlo.so
Scheduler::CallbackTaskScheduling()
vcl/source/app/scheduler.cxx:486 (discriminator 7)
program/libmergedlo.so
comphelper::SolarMutex::release(bool)
include/comphelper/solarmutex.hxx:91
program/libmergedlo.so
SvpSalInstance::ImplYield(bool, bool)
vcl/headless/svpinst.cxx:401
program/libmergedlo.so
SvpSalInstance::DoYield(bool, bool)
vcl/headless/svpinst.cxx:475
I.e. we do an abort when a task throws when the scheduler processes
events on the main loop. This happens *sometimes* locally, but it's
unclear to me how to reliably trigger it. When it happens, we warn like
this:
warn:vcl.schedule:27650:27441:vcl/source/app/scheduler.cxx:485: Uncaught
com.sun.star.lang.NotInitializedException message: "at
sfx2/source/doc/sfxbasemodel.cxx:3034" context: SdXImpressDocument
Checking the SAL_DEBUG_BACKTRACE() output for the case when
SfxBaseModel::MethodEntryCheck() throws:
#3 instdir/program/libsfxlo.so(SfxBaseModel::MethodEntryCheck(bool)
const+0x154) [0x7fec1e8ad9b4]
#4 instdir/program/libsfxlo.so(+0x6bdd1e) [0x7fec1e8bdd1e]
#5
instdir/program/libsfxlo.so(SfxBaseModel::getArgs2(com::sun::star::uno::Sequence<rtl::OUString>
const&)+0x56) [0x7fec1e8a0dd2]
#6 instdir/program/libsfxlo.so(SfxObjectShell::isEditDocLocked()
const+0x11c) [0x7fec1e8239da]
#7 instdir/program/libsfxlo.so(+0x882851) [0x7fec1ea82851]
#8 instdir/program/libsfxlo.so(+0x87cefc) [0x7fec1ea7cefc]
#9
instdir/program/libsfxlo.so(SfxDispatcher::FillState_(SfxSlotServer const&,
SfxItemSet&, SfxSlot const*)+0x1cc) [0x7fec1e4d57f4]
#10 instdir/program/libsfxlo.so(+0x2b900a) [0x7fec1e4b900a]
#11 instdir/program/libsfxlo.so(+0x2bd954) [0x7fec1e4bd954]
#12 instdir/program/libsfxlo.so(+0x2bd631) [0x7fec1e4bd631]
#13 instdir/program/libsfxlo.so(+0x2bd5a1) [0x7fec1e4bd5a1]
#14 instdir/program/libvcllo.so(+0xf4fb91) [0x7fec1894fb91]
#15 instdir/program/libvcllo.so(Timer::Invoke()+0x23)
[0x7fec1894fa19]
#16
instdir/program/libvcllo.so(Scheduler::CallbackTaskScheduling()+0x123d)
[0x7fec188fe277]
Fix this in SfxObjectShell::isEditDocLocked(), where we already return
false to avoid a nullptr dereference, and do it the way
SfxObjectShell::isScriptAccessAllowed() does the same for consistency.
Change-Id: I435b2877b1b600e01bf32b44bfa664be1672c231
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/180663
Reviewed-by: Miklos Vajna <vmik...@collabora.com>
Tested-by: Jenkins
diff --git a/sfx2/source/doc/objmisc.cxx b/sfx2/source/doc/objmisc.cxx
index a8005cab97e0..7d48d40a7439 100644
--- a/sfx2/source/doc/objmisc.cxx
+++ b/sfx2/source/doc/objmisc.cxx
@@ -2019,7 +2019,15 @@ bool SfxObjectShell::isEditDocLocked() const
if (officecfg::Office::Common::Misc::ViewerAppMode::get()
|| !officecfg::Office::Common::Misc::AllowEditReadonlyDocs::get())
return true;
- return comphelper::NamedValueCollection::getOrDefault(xModel->getArgs2( {
u"LockEditDoc"_ustr } ), u"LockEditDoc", false);
+ try
+ {
+ return
comphelper::NamedValueCollection::getOrDefault(xModel->getArgs2( {
u"LockEditDoc"_ustr } ), u"LockEditDoc", false);
+ }
+ catch (const uno::RuntimeException&)
+ {
+ TOOLS_WARN_EXCEPTION("sfx.appl", "unexpected RuntimeException");
+ }
+ return false;
}
bool SfxObjectShell::isContentExtractionLocked() const
