On Sun, Jul 14, 2024 at 5:13 PM Samuel Thibault <samuel.thiba...@gnu.org>
wrote:
> Samuel Thibault, le dim. 14 juil. 2024 18:11:40 +0200, a ecrit:
> > Samuel Thibault, le dim. 14 juil. 2024 18:06:22 +0200, a ecrit:
> > > Samuel Thibault, le dim. 14 juil. 2024 17:42:51 +0200, a ecrit:
> > > > Flavio Cruz, le dim. 14 juil. 2024 16:40:24 +0100, a ecrit:
> > > > > On 14.07.2024 16:37, Samuel Thibault wrote:
> > > > > > For inlined port arrays, WriteExtractArg compacts them from a
> > > > > > mach_port_name_inlined_t array to a mach_port_t array, reusing
> the
> > > > > > memory area. But when the server returns an error, the caller
> will
> > > > > > destroy the message, and thus expects the original inlined port
> arrays
> > > > > > available.
> > > > > >
> > > > > > Index: mig-1.8+git20231217/server.c
> > > > > >
> ===================================================================
> > > > > > --- mig-1.8+git20231217.orig/server.c
> > > > > > +++ mig-1.8+git20231217/server.c
> > > > > > @@ -790,6 +790,27 @@ WriteExtractArg(FILE *file, const argume
> > > > > > }
> > > > > >
> > > > > > static void
> > > > > > +WriteRestoreArg(FILE *file, const argument_t *arg)
> > > > > > +{
> > > > > > + if (akCheckAll(arg->argKind, akbSendRcv|akbPointer)) {
> > > > > > + if (akCheck(arg->argKind, akbIndefinite)) {
> > > > > > + fprintf(file, "\tif (OutP->%s != KERN_SUCCESS &&
> In%dP->%s%s.msgt_inline) {\n",
> > > > > > + arg->argRoutine->rtRetCode->argMsgField,
> > > > > > + arg->argRequestPos, arg->argTTName,
> arg->argLongForm ? ".msgtl_header" : "");
> > > > > > + fprintf(file, "\t\tmach_msg_type_number_t i;\n");
> > > > > > + fprintf(file, "\t\t/* Restore the
> mach_port_name_inlined_t input array for message destruction. */\n");
> > > > > > + fprintf(file, "\t\tfor (i = In%dP->%s.msgt%s_number -
> 1; i > 0; i--) {\n",
> > > > > > + arg->argRequestPos, arg->argTTName,
> arg->argLongForm ? "l" : "");
> > > > > > + fprintf(file, "\t\t\t%s[i].name = %sP[i];\n",
> InArgMsgField(arg), arg->argVarName);
> > > > > > + fprintf(file, "\t\t}\n");
> > > > > > + fprintf(file, "\t}\n");
> > > > > > + }
> > > > > > + else
> > > > > > + assert(false);
> > > > > > + }
> > > > > > +}
> > > > > > +
> > > > > > +static void
> > > > > > WriteServerCallArg(FILE *file, const argument_t *arg)
> > > > > > {
> > > > > > const ipc_type_t *it = arg->argType;
> > > > > > @@ -1436,6 +1457,8 @@ WriteRoutine(FILE *file, const routine_t
> > > > > > WriteServerCall(file, rt);
> > > > > > WriteGetReturnValue(file, rt);
> > > > > >
> > > > > > + WriteReverseList(file, rt->rtArgs, WriteRestoreArg,
> akbNone, "", "");
> > > > > > +
> > > > > > WriteReverseList(file, rt->rtArgs, WriteDestroyArg,
> akbDestroy, "", "");
> > > > > >
> > > > > > /*
> > > > >
> > > > > Tested the patch with cross-hurd and the change LGTM. Thanks for
> catching!
> > > >
> > > > I'm however seeing some issues that makes ext2fs hang when connecting
> > > > through ssh...
> > >
> > > It's the Restoration of other_handles in auth's authServer.c's
> > > _Xauth_makeauth stub that poses problem. If I comment it, I'm not
> > > getting the issue... Any idea out of lucky mind?
> >
> > Ah, perhaps nauth is 0 and thus In0P->other_handlesType.msgtl_number - 1
> > gets bogus, let me check.
>
> Yes, that's it, I'll shift the loop by 1
>
Ah yeah. That's why I do for (i = portsCnt; i > 0; --i) { ... } for server
output params to avoid subtracting the unsigned natural_t.
>
> Thanks for the review!
>
> Samuel
>
