1From 6cdb7b372b4d46e06a982c6f3494c086d2418c7b Mon Sep 17 00:00:00 2001
2From: Mark Adler <madler@alumni.caltech.edu>
3Date: Fri, 31 Jan 2020 22:05:59 -0800
4Subject: [PATCH] Detect and reject a zip bomb using overlapped entries.
5
6 Detect and reject a zip bomb using overlapped entries.
7
8 This detects an invalid zip file that has at least one entry that
9 overlaps with another entry or with the central directory to the
10 end of the file. A Fifield zip bomb uses overlapped local entries
11 to vastly increase the potential inflation ratio. Such an invalid
12 zip file is rejected.
13
14 See https://www.bamsoftware.com/hacks/zipbomb/ for David Fifield's
15 analysis, construction, and examples of such zip bombs.
16
17 The detection maintains a list of covered spans of the zip files
18 so far, where the central directory to the end of the file and any
19 bytes preceding the first entry at zip file offset zero are
20 considered covered initially. Then as each entry is decompressed
21 or tested, it is considered covered. When a new entry is about to
22 be processed, its initial offset is checked to see if it is
23 contained by a covered span. If so, the zip file is rejected as
24 invalid.
25
26 This commit depends on a preceding commit: "Fix bug in
27 undefer_input() that misplaced the input state."
28---
29 extract.c | 190 +++++++++++++++++++++++++++++++++++++++++++++++++++++-
30 globals.c | 1 +
31 globals.h | 3 +
32 process.c | 11 ++++
33 unzip.h | 1 +
34 5 files changed, 205 insertions(+), 1 deletion(-)
35
36diff --git a/extract.c b/extract.c
37index 549a5eb..1f078d1 100644
38--- a/extract.c
39+++ b/extract.c
40@@ -321,6 +321,125 @@ static ZCONST char Far UnsupportedExtraField[] =
41 "\nerror: unsupported extra-field compression type (%u)--skipping\n";
42 static ZCONST char Far BadExtraFieldCRC[] =
43 "error [%s]: bad extra-field CRC %08lx (should be %08lx)\n";
44+static ZCONST char Far NotEnoughMemCover[] =
45+ "error: not enough memory for bomb detection\n";
46+static ZCONST char Far OverlappedComponents[] =
47+ "error: invalid zip file with overlapped components (possible zip bomb)\n";
48+
49+
50+
51+
52+
53+/* A growable list of spans. */
54+typedef zoff_t bound_t;
55+typedef struct {
56+ bound_t beg; /* start of the span */
57+ bound_t end; /* one past the end of the span */
58+} span_t;
59+typedef struct {
60+ span_t *span; /* allocated, distinct, and sorted list of spans */
61+ size_t num; /* number of spans in the list */
62+ size_t max; /* allocated number of spans (num <= max) */
63+} cover_t;
64+
65+/*
66+ * Return the index of the first span in cover whose beg is greater than val.
67+ * If there is no such span, then cover->num is returned.
68+ */
69+static size_t cover_find(cover, val)
70+ cover_t *cover;
71+ bound_t val;
72+{
73+ size_t lo = 0, hi = cover->num;
74+ while (lo < hi) {
75+ size_t mid = (lo + hi) >> 1;
76+ if (val < cover->span[mid].beg)
77+ hi = mid;
78+ else
79+ lo = mid + 1;
80+ }
81+ return hi;
82+}
83+
84+/* Return true if val lies within any one of the spans in cover. */
85+static int cover_within(cover, val)
86+ cover_t *cover;
87+ bound_t val;
88+{
89+ size_t pos = cover_find(cover, val);
90+ return pos > 0 && val < cover->span[pos - 1].end;
91+}
92+
93+/*
94+ * Add a new span to the list, but only if the new span does not overlap any
95+ * spans already in the list. The new span covers the values beg..end-1. beg
96+ * must be less than end.
97+ *
98+ * Keep the list sorted and merge adjacent spans. Grow the allocated space for
99+ * the list as needed. On success, 0 is returned. If the new span overlaps any
100+ * existing spans, then 1 is returned and the new span is not added to the
101+ * list. If the new span is invalid because beg is greater than or equal to
102+ * end, then -1 is returned. If the list needs to be grown but the memory
103+ * allocation fails, then -2 is returned.
104+ */
105+static int cover_add(cover, beg, end)
106+ cover_t *cover;
107+ bound_t beg;
108+ bound_t end;
109+{
110+ size_t pos;
111+ int prec, foll;
112+
113+ if (beg >= end)
114+ /* The new span is invalid. */
115+ return -1;
116+
117+ /* Find where the new span should go, and make sure that it does not
118+ overlap with any existing spans. */
119+ pos = cover_find(cover, beg);
120+ if ((pos > 0 && beg < cover->span[pos - 1].end) ||
121+ (pos < cover->num && end > cover->span[pos].beg))
122+ return 1;
123+
124+ /* Check for adjacencies. */
125+ prec = pos > 0 && beg == cover->span[pos - 1].end;
126+ foll = pos < cover->num && end == cover->span[pos].beg;
127+ if (prec && foll) {
128+ /* The new span connects the preceding and following spans. Merge the
129+ following span into the preceding span, and delete the following
130+ span. */
131+ cover->span[pos - 1].end = cover->span[pos].end;
132+ cover->num--;
133+ memmove(cover->span + pos, cover->span + pos + 1,
134+ (cover->num - pos) * sizeof(span_t));
135+ }
136+ else if (prec)
137+ /* The new span is adjacent only to the preceding span. Extend the end
138+ of the preceding span. */
139+ cover->span[pos - 1].end = end;
140+ else if (foll)
141+ /* The new span is adjacent only to the following span. Extend the
142+ beginning of the following span. */
143+ cover->span[pos].beg = beg;
144+ else {
145+ /* The new span has gaps between both the preceding and the following
146+ spans. Assure that there is room and insert the span. */
147+ if (cover->num == cover->max) {
148+ size_t max = cover->max == 0 ? 16 : cover->max << 1;
149+ span_t *span = realloc(cover->span, max * sizeof(span_t));
150+ if (span == NULL)
151+ return -2;
152+ cover->span = span;
153+ cover->max = max;
154+ }
155+ memmove(cover->span + pos + 1, cover->span + pos,
156+ (cover->num - pos) * sizeof(span_t));
157+ cover->num++;
158+ cover->span[pos].beg = beg;
159+ cover->span[pos].end = end;
160+ }
161+ return 0;
162+}
163
164
165
166@@ -376,6 +495,29 @@ int extract_or_test_files(__G) /* return PK-type error code */
167 }
168 #endif /* !SFX || SFX_EXDIR */
169
170+ /* One more: initialize cover structure for bomb detection. Start with a
171+ span that covers the central directory though the end of the file. */
172+ if (G.cover == NULL) {
173+ G.cover = malloc(sizeof(cover_t));
174+ if (G.cover == NULL) {
175+ Info(slide, 0x401, ((char *)slide,
176+ LoadFarString(NotEnoughMemCover)));
177+ return PK_MEM;
178+ }
179+ ((cover_t *)G.cover)->span = NULL;
180+ ((cover_t *)G.cover)->max = 0;
181+ }
182+ ((cover_t *)G.cover)->num = 0;
183+ if ((G.extra_bytes != 0 &&
184+ cover_add((cover_t *)G.cover, 0, G.extra_bytes) != 0) ||
185+ cover_add((cover_t *)G.cover,
186+ G.extra_bytes + G.ecrec.offset_start_central_directory,
187+ G.ziplen) != 0) {
188+ Info(slide, 0x401, ((char *)slide,
189+ LoadFarString(NotEnoughMemCover)));
190+ return PK_MEM;
191+ }
192+
193 /*---------------------------------------------------------------------------
194 The basic idea of this function is as follows. Since the central di-
195 rectory lies at the end of the zipfile and the member files lie at the
196@@ -593,7 +735,8 @@ int extract_or_test_files(__G) /* return PK-type error code */
197 if (error > error_in_archive)
198 error_in_archive = error;
199 /* ...and keep going (unless disk full or user break) */
200- if (G.disk_full > 1 || error_in_archive == IZ_CTRLC) {
201+ if (G.disk_full > 1 || error_in_archive == IZ_CTRLC ||
202+ error == PK_BOMB) {
203 /* clear reached_end to signal premature stop ... */
204 reached_end = FALSE;
205 /* ... and cancel scanning the central directory */
206@@ -1062,6 +1205,11 @@ static int extract_or_test_entrylist(__G__ numchunk,
207
208 /* seek_zipf(__G__ pInfo->offset); */
209 request = G.pInfo->offset + G.extra_bytes;
210+ if (cover_within((cover_t *)G.cover, request)) {
211+ Info(slide, 0x401, ((char *)slide,
212+ LoadFarString(OverlappedComponents)));
213+ return PK_BOMB;
214+ }
215 inbuf_offset = request % INBUFSIZ;
216 bufstart = request - inbuf_offset;
217
218@@ -1602,6 +1750,18 @@ reprompt:
219 return IZ_CTRLC; /* cancel operation by user request */
220 }
221 #endif
222+ error = cover_add((cover_t *)G.cover, request,
223+ G.cur_zipfile_bufstart + (G.inptr - G.inbuf));
224+ if (error < 0) {
225+ Info(slide, 0x401, ((char *)slide,
226+ LoadFarString(NotEnoughMemCover)));
227+ return PK_MEM;
228+ }
229+ if (error != 0) {
230+ Info(slide, 0x401, ((char *)slide,
231+ LoadFarString(OverlappedComponents)));
232+ return PK_BOMB;
233+ }
234 #ifdef MACOS /* MacOS is no preemptive OS, thus call event-handling by hand */
235 UserStop();
236 #endif
237@@ -2003,6 +2163,34 @@ static int extract_or_test_member(__G) /* return PK-type error code */
238 }
239
240 undefer_input(__G);
241+
242+ if ((G.lrec.general_purpose_bit_flag & 8) != 0) {
243+ /* skip over data descriptor (harder than it sounds, due to signature
244+ * ambiguity)
245+ */
246+# define SIG 0x08074b50
247+# define LOW 0xffffffff
248+ uch buf[12];
249+ unsigned shy = 12 - readbuf((char *)buf, 12);
250+ ulg crc = shy ? 0 : makelong(buf);
251+ ulg clen = shy ? 0 : makelong(buf + 4);
252+ ulg ulen = shy ? 0 : makelong(buf + 8); /* or high clen if ZIP64 */
253+ if (crc == SIG && /* if not SIG, no signature */
254+ (G.lrec.crc32 != SIG || /* if not SIG, have signature */
255+ (clen == SIG && /* if not SIG, no signature */
256+ ((G.lrec.csize & LOW) != SIG || /* if not SIG, have signature */
257+ (ulen == SIG && /* if not SIG, no signature */
258+ (G.zip64 ? G.lrec.csize >> 32 : G.lrec.ucsize) != SIG
259+ /* if not SIG, have signature */
260+ )))))
261+ /* skip four more bytes to account for signature */
262+ shy += 4 - readbuf((char *)buf, 4);
263+ if (G.zip64)
264+ shy += 8 - readbuf((char *)buf, 8); /* skip eight more for ZIP64 */
265+ if (shy)
266+ error = PK_ERR;
267+ }
268+
269 return error;
270
271 } /* end function extract_or_test_member() */
272diff --git a/globals.c b/globals.c
273index fa8cca5..1e0f608 100644
274--- a/globals.c
275+++ b/globals.c
276@@ -181,6 +181,7 @@ Uz_Globs *globalsCtor()
277 # if (!defined(NO_TIMESTAMPS))
278 uO.D_flag=1; /* default to '-D', no restoration of dir timestamps */
279 # endif
280+ G.cover = NULL; /* not allocated yet */
281 #endif
282
283 uO.lflag=(-1);
284diff --git a/globals.h b/globals.h
285index 11b7215..2bdcdeb 100644
286--- a/globals.h
287+++ b/globals.h
288@@ -260,12 +260,15 @@ typedef struct Globals {
289 ecdir_rec ecrec; /* used in unzip.c, extract.c */
290 z_stat statbuf; /* used by main, mapname, check_for_newer */
291
292+ int zip64; /* true if Zip64 info in extra field */
293+
294 int mem_mode;
295 uch *outbufptr; /* extract.c static */
296 ulg outsize; /* extract.c static */
297 int reported_backslash; /* extract.c static */
298 int disk_full;
299 int newfile;
300+ void **cover; /* used in extract.c for bomb detection */
301
302 int didCRlast; /* fileio static */
303 ulg numlines; /* fileio static: number of lines printed */
304diff --git a/process.c b/process.c
305index e4f2405..e4e7aee 100644
306--- a/process.c
307+++ b/process.c
308@@ -637,6 +637,13 @@ void free_G_buffers(__G) /* releases all memory allocated in global vars */
309 }
310 #endif
311
312+ /* Free the cover span list and the cover structure. */
313+ if (G.cover != NULL) {
314+ free(*(G.cover));
315+ free(G.cover);
316+ G.cover = NULL;
317+ }
318+
319 } /* end function free_G_buffers() */
320
321
322@@ -1913,6 +1920,8 @@ int getZip64Data(__G__ ef_buf, ef_len)
323 #define Z64FLGS 0xffff
324 #define Z64FLGL 0xffffffff
325
326+ G.zip64 = FALSE;
327+
328 if (ef_len == 0 || ef_buf == NULL)
329 return PK_COOL;
330
331@@ -2084,6 +2093,8 @@ int getUnicodeData(__G__ ef_buf, ef_len)
332 (ZCONST char *)(offset + ef_buf), ULen);
333 G.unipath_filename[ULen] = '\0';
334 }
335+
336+ G.zip64 = TRUE;
337 }
338
339 /* Skip this extra field block */
340diff --git a/unzip.h b/unzip.h
341index 5b2a326..ed24a5b 100644
342--- a/unzip.h
343+++ b/unzip.h
344@@ -645,6 +645,7 @@ typedef struct _Uzp_cdir_Rec {
345 #define PK_NOZIP 9 /* zipfile not found */
346 #define PK_PARAM 10 /* bad or illegal parameters specified */
347 #define PK_FIND 11 /* no files found */
348+#define PK_BOMB 12 /* likely zip bomb */
349 #define PK_DISK 50 /* disk full */
350 #define PK_EOF 51 /* unexpected EOF */
351
352--
3532.25.0
354
