From 0287c18a0d7b5244eab93aa35ca8d71ddcf10f22 Mon Sep 17 00:00:00 2001
From: Demi Marie Obenour <demi@invisiblethingslab.com>
Date: Sun, 25 Dec 2022 14:21:31 -0500
Subject: [PATCH] Reject all unassigned and invalid Unicode codepoints

This required build system changes, as the resulting generated table
really needs to be in a separate file.
---
 qrexec-lib/Makefile               | 15 ++++++++++---
 gentbl.py => qrexec-lib/gentbl.py |  4 +---
 qrexec-lib/unpack.c               | 37 +------------------------------
 3 files changed, 14 insertions(+), 42 deletions(-)
 rename gentbl.py => qrexec-lib/gentbl.py (90%)

diff --git a/qrexec-lib/Makefile b/qrexec-lib/Makefile
index de395713..5053fe6f 100644
--- a/qrexec-lib/Makefile
+++ b/qrexec-lib/Makefile
@@ -2,12 +2,20 @@ CC=gcc
 CFLAGS+=-I. -g -O2 -Wall -Wextra -Werror -pie -fPIC
 SO_VER=2
 LDFLAGS+=-shared
+.PHONY: all clean install
+objs := ioall.o copy-file.o crc32.o unpack.o pack.o
 
 all: libqubes-rpc-filecopy.so.$(SO_VER)
-libqubes-rpc-filecopy.so.$(SO_VER): ioall.o copy-file.o crc32.o unpack.o pack.o
-	$(CC) $(LDFLAGS) -Wl,-soname,$@ -o $@ $^
+libqubes-rpc-filecopy.so.$(SO_VER): $(objs)
+	$(CC) $(LDFLAGS) -Wl,-soname,$@ -o $@ $^ $(LDLIBS)
 
-%.a:
+%.o: %.c
+	$(CC) $(CFLAGS) -MD -MP -MF $@.dep -c -o $@ $<
+unpack.o: unpack-table.c
+unpack-table.c: gentbl.py
+	python3 gentbl.py > unpack-table.c
+
+%.a: $(objs)
 	$(AR) rcs $@ $^
 clean:
 	rm -f *.o *~ *.a *.so.*
@@ -18,3 +26,4 @@ install:
 	ln -s libqubes-rpc-filecopy.so.$(SO_VER) $(DESTDIR)$(LIBDIR)/libqubes-rpc-filecopy.so
 	mkdir -p $(DESTDIR)$(INCLUDEDIR)
 	cp libqubes-rpc-filecopy.h $(DESTDIR)$(INCLUDEDIR)
+-include ./*.o.dep
diff --git a/gentbl.py b/qrexec-lib/gentbl.py
similarity index 90%
rename from gentbl.py
rename to qrexec-lib/gentbl.py
index 96194d70..ba3e227a 100755
--- a/gentbl.py
+++ b/qrexec-lib/gentbl.py
@@ -2,8 +2,6 @@
 import sys
 def main():
     def print_interval(interval, last_cat):
-        if last_cat == 'Cn':
-            return
         if interval[0] != interval[1]:
             print(f'    case 0x{interval[0]:X} ... 0x{interval[1]:X}: // category {last_cat}'
                   .replace('category Cs', 'surrogates'))
@@ -25,7 +23,7 @@ def print_interval(interval, last_cat):
                 print_interval(interval, last_cat)
             interval = [i, i]
     print_interval(interval, last_cat)
-    print('    case 0x10FFFF ... UINT32_MAX: // too large')
+    print('    case 0x110000 ... UINT32_MAX: // too large')
     sys.stdout.flush()
 if __name__ == '__main__':
     main()
diff --git a/qrexec-lib/unpack.c b/qrexec-lib/unpack.c
index 8d5bfd5e..1fc666b6 100644
--- a/qrexec-lib/unpack.c
+++ b/qrexec-lib/unpack.c
@@ -242,43 +242,8 @@ static int validate_utf8_char(const unsigned char *untrusted_c) {
     }
 
     switch (code_point) {
-    case 0x0 ... 0x1F: // category Cc
-    case 0x7F ... 0x9F: // category Cc
-    case 0xA0: // category Zs
-    case 0xAD: // category Cf
-    case 0x600 ... 0x605: // category Cf
-    case 0x61C: // category Cf
-    case 0x6DD: // category Cf
-    case 0x70F: // category Cf
-    case 0x8E2: // category Cf
-    case 0x1680: // category Zs
-    case 0x180E: // category Cf
-    case 0x2000 ... 0x200A: // category Zs
-    case 0x200B ... 0x200F: // category Cf
-    case 0x2028: // category Zl
-    case 0x2029: // category Zp
-    case 0x202A ... 0x202E: // category Cf
-    case 0x202F: // category Zs
-    case 0x205F: // category Zs
-    case 0x2060 ... 0x2064: // category Cf
-    case 0x2066 ... 0x206F: // category Cf
-    case 0x3000: // category Zs
-    case 0xD800 ... 0xDFFF: // surrogates
-    case 0xE000 ... 0xF8FF: // category Co
-    case 0xFEFF: // category Cf
-    case 0xFFF9 ... 0xFFFB: // category Cf
-    case 0x110BD: // category Cf
-    case 0x110CD: // category Cf
-    case 0x13430 ... 0x13438: // category Cf
-    case 0x1BCA0 ... 0x1BCA3: // category Cf
-    case 0x1D173 ... 0x1D17A: // category Cf
-    case 0xE0001: // category Cf
-    case 0xE0020 ... 0xE007F: // category Cf
-    case 0xF0000 ... 0xFFFFD: // category Co
-    case 0x100000 ... 0x10FFFD: // category Co
-    case 0x110000 ... UINT32_MAX: // too large
+#include "unpack-table.c"
         return 0; // Invalid UTF-8 or forbidden codepoint
-
     default:
         return total_size;
     }