Script 'mail_helper' called by obssrc Hello community, here is the log from the commit of package libXfont2 for openSUSE:Factory checked in at 2026-11-19 16:53:39 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ Comparing /work/SRC/openSUSE:Factory/libXfont2 (Old) and /work/SRC/openSUSE:Factory/.libXfont2.new.1251 (New) ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Package is "libXfont2" Thu Nov 19 16:53:39 2026 rev:11 rq:1398279 version:2.0.7 Changes: -------- --- /work/SRC/openSUSE:Factory/libXfont2/libXfont2.changes 2026-08-06 16:19:04.637904563 +0200 +++ /work/SRC/openSUSE:Factory/.libXfont2.new.1251/libXfont2.changes 2026-11-19 16:53:49.732511979 +0100 @@ -1,0 +2,14 @@ +Tue Nov 17 10:32:39 UTC 2026 - Stefan Dirsch <[email protected]> + +- bsc1289838_CVE-2026-109920_FreeType-always-initialise-strike_index.patch + * FreeType backend uninitialised strike_index out-of-bounds read + (CVE-2026-109920, bsc#1289838) + +------------------------------------------------------------------- +Thu Nov 5 13:21:15 UTC 2026 - Stefan Dirsch <[email protected]> + +- bsc1289304_CVE-2026-109909_bufio-pcf-reject-negative-skip-count-in-BufFileRawSk.patch + * PCF Font Parsing Negative Skip Out-Of-Bounds Read (CVE-2026-109909, + bsc#1289304) + +------------------------------------------------------------------- New: ---- bsc1289304_CVE-2026-109909_bufio-pcf-reject-negative-skip-count-in-BufFileRawSk.patch bsc1289838_CVE-2026-109920_FreeType-always-initialise-strike_index.patch ----------(New B)---------- New: - bsc1289304_CVE-2026-109909_bufio-pcf-reject-negative-skip-count-in-BufFileRawSk.patch * PCF Font Parsing Negative Skip Out-Of-Bounds Read (CVE-2026-109909, New: - bsc1289838_CVE-2026-109920_FreeType-always-initialise-strike_index.patch * FreeType backend uninitialised strike_index out-of-bounds read ----------(New E)---------- ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ Other differences: ------------------ ++++++ libXfont2.spec ++++++ --- /var/tmp/diff_new_pack.MLbqA2/_old 2026-11-19 16:53:50.692552673 +0100 +++ /var/tmp/diff_new_pack.MLbqA2/_new 2026-11-19 16:53:50.693552716 +0100 @@ -34,6 +34,8 @@ Patch1269020: bsc1269020_CVE-2026-56003_0003-bitscale-add-bounds-check-to-computeProps-for-proper.patch Patch1272660: bsc1272660_CVE-2026-59679_0001-fserve-validate-num_chars-against-encoding-array-siz.patch Patch1272661: bsc1272661_CVE-2026-44950-0002-fserve-bounds-check-cumulative-glyph-data-writes-in-.patch +Patch1289304: bsc1289304_CVE-2026-109909_bufio-pcf-reject-negative-skip-count-in-BufFileRawSk.patch +Patch1289838: bsc1289838_CVE-2026-109920_FreeType-always-initialise-strike_index.patch BuildRoot: %{_tmppath}/%{name}-%{version}-build #git#BuildRequires: autoconf >= 2.60, automake, libtool BuildRequires: pkgconfig ++++++ bsc1289304_CVE-2026-109909_bufio-pcf-reject-negative-skip-count-in-BufFileRawSk.patch ++++++ >From f9f25c0c5d1f60ebcdcf9f83fa57bf0ec6cc9a20 Mon Sep 17 00:00:00 2001 From: Peter Hutterer <[email protected]> Date: Tue, 3 Nov 2026 15:53:12 +1000 Subject: [PATCH libXfont] bufio,pcf: reject negative skip count in BufFileRawSkip MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit BufFileRawSkip() takes a signed int count but never validates it for negative values. A crafted PCF font file can have a table offset that, when subtracted from the current file position, wraps to a negative value. This negative count passes the buffer bounds check (curoff + count <= fileoff is trivially true for negative count) and moves the buffer pointer backward past the buffer start. Subsequent FontFileGetc() calls then read from invalid memory, causing an out-of-bounds heap read or a segmentation fault. Add validation in BufFileRawSkip() to reject negative count values, returning BUFFILEEOF. Additionally, add defense-in-depth checks in pcfSeekToType() to validate that the skip distance fits in an int and that the position is non-negative before calling FontFileSkip. This vulnerability was discovered by: Elias Myllymäki working with TrendAI Zero Day Initiative CVE-2026-109909, ZDI-CAN-33183 Assisted-by: AI --- src/bitmap/pcfread.c | 4 +++- src/fontfile/bufio.c | 3 +++ 2 files changed, 6 insertions(+), 1 deletion(-) diff --git ./src/bitmap/pcfread.c ../src/bitmap/pcfread.c index a385331e1d48..e3b6277e1ed5 100644 --- ./src/bitmap/pcfread.c +++ ../src/bitmap/pcfread.c @@ -203,17 +203,19 @@ pcfGetCompressedMetric(FontFilePtr file, CARD32 format, xCharInfo *metric) static Bool pcfSeekToType(FontFilePtr file, PCFTablePtr tables, int ntables, CARD32 type, CARD32 *formatp, CARD32 *sizep) { int i; for (i = 0; i < ntables; i++) if (tables[i].type == type) { - if (position > tables[i].offset) + if (position < 0 || (CARD32)position > tables[i].offset) + return FALSE; + if (tables[i].offset - (CARD32)position > INT_MAX) return FALSE; if (!FontFileSkip(file, tables[i].offset - position)) return FALSE; position = tables[i].offset; *sizep = tables[i].size; *formatp = tables[i].format; return TRUE; } diff --git ./src/fontfile/bufio.c ../src/fontfile/bufio.c index 7465f72cbfe3..8b081e9bf595 100644 --- ./src/fontfile/bufio.c +++ ../src/fontfile/bufio.c @@ -82,16 +82,19 @@ BufFileRawFill (BufFilePtr f) static int BufFileRawSkip (BufFilePtr f, int count) { off_t curoff; off_t fileoff; off_t todo; + if (count < 0) + return BUFFILEEOF; + curoff = f->bufp - f->buffer; fileoff = curoff + f->left; if (curoff + count <= fileoff) { f->bufp += count; f->left -= count; } else { todo = count - (fileoff - curoff); if (lseek (FileDes(f), todo, SEEK_CUR) == -1) { -- 2.55.0 ++++++ bsc1289838_CVE-2026-109920_FreeType-always-initialise-strike_index.patch ++++++ >From 0ccab39c617c328b4b126a033656e9a9ece44772 Mon Sep 17 00:00:00 2001 From: Peter Hutterer <[email protected]> Date: Mon, 9 Nov 2026 14:37:01 +1000 Subject: [PATCH libXfont 2/2] FreeType: always initialise strike_index When FreeTypeOpenInstance() allocates FTInstanceRec, the instance->strike_index field is the only one left unset by default. By renaming a BDF file to a FreeType-backed .pfb file we can trick libXfont2 to use the FreeType backend, not its own bitmap backend, resulting in a "non-SFNT face". This leaves the strike_index field unset. Later when we call FT_Do_SBit_Metrics() strike_index is used directly as an array subscript, leaving us with potential OOB reads: FT_Bitmap_Size* sz = &ft_face->available_sizes[strike_index]; This vulnerability was discovered by: Apex Security Research ([email protected]) CVE-2026-109920 Assisted-by: Claude:claude-opus-4-6 --- Makefile.am | 12 ++ src/FreeType/ftfuncs.c | 6 +- test/data/ft-strike-index/fonts.dir | 2 + test/data/ft-strike-index/strike-test.bdf | 52 ++++++++ test/data/ft-strike-index/strike-test.pfb | 52 ++++++++ test/test-freetype-strike-index.c | 154 ++++++++++++++++++++++ 6 files changed, 275 insertions(+), 3 deletions(-) create mode 100644 test/data/ft-strike-index/fonts.dir create mode 100644 test/data/ft-strike-index/strike-test.bdf create mode 100644 test/data/ft-strike-index/strike-test.pfb create mode 100644 test/test-freetype-strike-index.c diff --git ./src/FreeType/ftfuncs.c ../src/FreeType/ftfuncs.c index 0d529a8cee5b..d6117b341c11 100644 --- ./src/FreeType/ftfuncs.c +++ ../src/FreeType/ftfuncs.c @@ -418,16 +418,17 @@ FreeTypeOpenInstance(FTInstancePtr *instance_return, FTFacePtr face, if(instance == NULL) { return AllocError; } instance->refcount = 1; instance->face = face; instance->load_flags = load_flags; + instance->strike_index = 0xFFFFU; instance->spacing = spacing; /* Actual spacing */ instance->pixel_size =0; instance->pixel_width_unit_x =0; instance->pixel_width_unit_y =0; instance->charcellMetrics = NULL; instance->averageWidth = 0; instance->rawAverageWidth = 0; instance->forceConstantMetrics = NULL; @@ -473,18 +474,16 @@ FreeTypeOpenInstance(FTInstancePtr *instance_return, FTFacePtr face, return FTtoXReturnCode(ftrc); } if( FT_IS_SFNT( face->face ) ) { #if 1 FT_F26Dot6 tt_char_width, tt_char_height, tt_dim_x, tt_dim_y; FT_Int nn; - instance->strike_index=0xFFFFU; - tt_char_width = (FT_F26Dot6)(trans->scale*(1<<6) + 0.5); tt_char_height = (FT_F26Dot6)(trans->scale*(1<<6) + 0.5); tt_dim_x = FLOOR64( ( tt_char_width * trans->xres + 36 ) / 72 + 32 ); tt_dim_y = FLOOR64( ( tt_char_height * trans->yres + 36 ) / 72 + 32 ); if ( tt_dim_x && !tt_dim_y ) tt_dim_y = tt_dim_x; @@ -969,17 +968,18 @@ ft_get_very_lazy_bbox( FT_UInt index, } static FT_Error FT_Do_SBit_Metrics( FT_Face ft_face, FT_Size ft_size, FT_ULong strike_index, FT_UShort glyph_index, FT_Glyph_Metrics *metrics_return, int *sbitchk_incomplete_but_exist ) { #if 1 - if ( strike_index != 0xFFFFU && ft_face->available_sizes != NULL ) + if ( strike_index != 0xFFFFU && ft_face->available_sizes != NULL && + strike_index < (FT_ULong) ft_face->num_fixed_sizes ) { FT_Error error; FT_Bitmap_Size* sz = &ft_face->available_sizes[strike_index]; error = FT_Set_Pixel_Sizes( ft_face, sz->x_ppem/64, sz->y_ppem/64 ); if ( !error ) { error = FT_Load_Glyph( ft_face, glyph_index, FT_LOAD_SBITS_ONLY ); -- 2.56.0
