| Message ID | 1529056145-6404-1-git-send-email-thuth@redhat.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Fri, 15 Jun 2018 11:49:05 +0200 Thomas Huth <thuth@redhat.com> wrote: > The rom_ptr() function allows direct access to the ROM blobs that we > load during startup. However, there are currently no checks for the > size of the accesses, so it's currently possible to crash QEMU for > example with: > > $ echo "Insane in the mainframe" > /tmp/test.txt > $ s390x-softmmu/qemu-system-s390x -kernel /tmp/test.txt -append xyz > Segmentation fault (core dumped) > $ s390x-softmmu/qemu-system-s390x -kernel /tmp/test.txt -initrd /tmp/test.txt > Segmentation fault (core dumped) > $ echo -n HdrS > /tmp/hdr.txt > $ sparc64-softmmu/qemu-system-sparc64 -kernel /tmp/hdr.txt -initrd /tmp/hdr.txt > Segmentation fault (core dumped) > > We need a possibility to check the size of the ROM area that we want > to access, thus let's add a size parameter to the rom_ptr() function > to avoid these problems. > > Signed-off-by: Thomas Huth <thuth@redhat.com> > --- > v3: Fix the check in find_rom() > > hw/core/loader.c | 10 +++++----- > hw/mips/mips_malta.c | 6 ++++-- > hw/s390x/ipl.c | 13 ++++++++++--- > hw/sparc/sun4m.c | 4 ++-- > hw/sparc64/sun4u.c | 4 ++-- > include/hw/loader.h | 2 +- > target/arm/cpu.c | 2 +- > 7 files changed, 25 insertions(+), 16 deletions(-) > diff --git a/hw/s390x/ipl.c b/hw/s390x/ipl.c > index 04245b5..bcac204 100644 > --- a/hw/s390x/ipl.c > +++ b/hw/s390x/ipl.c > @@ -195,8 +199,11 @@ static void s390_ipl_realize(DeviceState *dev, Error **errp) > * we have to overwrite values in the kernel image, > * which are "rom" > */ > - stq_p(rom_ptr(INITRD_PARM_START), initrd_offset); > - stq_p(rom_ptr(INITRD_PARM_SIZE), initrd_size); > + romptr = rom_ptr(INITRD_PARM_START, 16); > + if (romptr) { > + stq_p(romptr, initrd_offset); > + stq_p(romptr + 1, initrd_size); INITRD_PARM_SIZE seems to be unused now. We can remove it in a separate patch if you don't respin, though. > + } > } > } > /* Reviewed-by: Cornelia Huck <cohuck@redhat.com>
On 15.06.2018 11:49, Thomas Huth wrote: > The rom_ptr() function allows direct access to the ROM blobs that we > load during startup. However, there are currently no checks for the > size of the accesses, so it's currently possible to crash QEMU for > example with: > > $ echo "Insane in the mainframe" > /tmp/test.txt > $ s390x-softmmu/qemu-system-s390x -kernel /tmp/test.txt -append xyz > Segmentation fault (core dumped) > $ s390x-softmmu/qemu-system-s390x -kernel /tmp/test.txt -initrd /tmp/test.txt > Segmentation fault (core dumped) > $ echo -n HdrS > /tmp/hdr.txt > $ sparc64-softmmu/qemu-system-sparc64 -kernel /tmp/hdr.txt -initrd /tmp/hdr.txt > Segmentation fault (core dumped) > > We need a possibility to check the size of the ROM area that we want > to access, thus let's add a size parameter to the rom_ptr() function > to avoid these problems. > > Signed-off-by: Thomas Huth <thuth@redhat.com> Ping! Could anybody please pick this patch up? Qemu-trivial seems to be pretty dormant these days (?), so maybe Paolo via misc? Or either the s390x or Sparc tree, since it fixes a crash on these machines? Thanks, Thomas
On 06/25/2018 10:31 AM, Thomas Huth wrote: > On 15.06.2018 11:49, Thomas Huth wrote: >> The rom_ptr() function allows direct access to the ROM blobs that we >> load during startup. However, there are currently no checks for the >> size of the accesses, so it's currently possible to crash QEMU for >> example with: >> >> $ echo "Insane in the mainframe" > /tmp/test.txt >> $ s390x-softmmu/qemu-system-s390x -kernel /tmp/test.txt -append xyz >> Segmentation fault (core dumped) >> $ s390x-softmmu/qemu-system-s390x -kernel /tmp/test.txt -initrd /tmp/test.txt >> Segmentation fault (core dumped) >> $ echo -n HdrS > /tmp/hdr.txt >> $ sparc64-softmmu/qemu-system-sparc64 -kernel /tmp/hdr.txt -initrd /tmp/hdr.txt >> Segmentation fault (core dumped) >> >> We need a possibility to check the size of the ROM area that we want >> to access, thus let's add a size parameter to the rom_ptr() function >> to avoid these problems. >> >> Signed-off-by: Thomas Huth <thuth@redhat.com> > > Ping! > > Could anybody please pick this patch up? Qemu-trivial seems to be pretty > dormant these days (?), so maybe Paolo via misc? Or either the s390x or > Sparc tree, since it fixes a crash on these machines? Acked-by: Christian Borntraeger <borntraeger@de.ibm.com> if this will go via the s390x tree.
On Mon, 25 Jun 2018 10:31:52 +0200 Thomas Huth <thuth@redhat.com> wrote: > On 15.06.2018 11:49, Thomas Huth wrote: > > The rom_ptr() function allows direct access to the ROM blobs that we > > load during startup. However, there are currently no checks for the > > size of the accesses, so it's currently possible to crash QEMU for > > example with: > > > > $ echo "Insane in the mainframe" > /tmp/test.txt > > $ s390x-softmmu/qemu-system-s390x -kernel /tmp/test.txt -append xyz > > Segmentation fault (core dumped) > > $ s390x-softmmu/qemu-system-s390x -kernel /tmp/test.txt -initrd /tmp/test.txt > > Segmentation fault (core dumped) > > $ echo -n HdrS > /tmp/hdr.txt > > $ sparc64-softmmu/qemu-system-sparc64 -kernel /tmp/hdr.txt -initrd /tmp/hdr.txt > > Segmentation fault (core dumped) > > > > We need a possibility to check the size of the ROM area that we want > > to access, thus let's add a size parameter to the rom_ptr() function > > to avoid these problems. > > > > Signed-off-by: Thomas Huth <thuth@redhat.com> > > Ping! > > Could anybody please pick this patch up? Qemu-trivial seems to be pretty > dormant these days (?), so maybe Paolo via misc? Or either the s390x or > Sparc tree, since it fixes a crash on these machines? If nobody else wants it, I can take it through the s390x tree. Would not mind some more acks, though.
On Fri, 15 Jun 2018 11:49:05 +0200 Thomas Huth <thuth@redhat.com> wrote: > The rom_ptr() function allows direct access to the ROM blobs that we > load during startup. However, there are currently no checks for the > size of the accesses, so it's currently possible to crash QEMU for > example with: > > $ echo "Insane in the mainframe" > /tmp/test.txt > $ s390x-softmmu/qemu-system-s390x -kernel /tmp/test.txt -append xyz > Segmentation fault (core dumped) > $ s390x-softmmu/qemu-system-s390x -kernel /tmp/test.txt -initrd /tmp/test.txt > Segmentation fault (core dumped) > $ echo -n HdrS > /tmp/hdr.txt > $ sparc64-softmmu/qemu-system-sparc64 -kernel /tmp/hdr.txt -initrd /tmp/hdr.txt > Segmentation fault (core dumped) > > We need a possibility to check the size of the ROM area that we want > to access, thus let's add a size parameter to the rom_ptr() function > to avoid these problems. > > Signed-off-by: Thomas Huth <thuth@redhat.com> Grmpf, this conflicts with "s390/ipl: fix ipl with -no-reboot" in s390-next... do you want to rebase it? > --- > v3: Fix the check in find_rom() > > hw/core/loader.c | 10 +++++----- > hw/mips/mips_malta.c | 6 ++++-- > hw/s390x/ipl.c | 13 ++++++++++--- > hw/sparc/sun4m.c | 4 ++-- > hw/sparc64/sun4u.c | 4 ++-- > include/hw/loader.h | 2 +- > target/arm/cpu.c | 2 +- > 7 files changed, 25 insertions(+), 16 deletions(-)
On 26.06.2018 10:40, Cornelia Huck wrote: > On Fri, 15 Jun 2018 11:49:05 +0200 > Thomas Huth <thuth@redhat.com> wrote: > >> The rom_ptr() function allows direct access to the ROM blobs that we >> load during startup. However, there are currently no checks for the >> size of the accesses, so it's currently possible to crash QEMU for >> example with: >> >> $ echo "Insane in the mainframe" > /tmp/test.txt >> $ s390x-softmmu/qemu-system-s390x -kernel /tmp/test.txt -append xyz >> Segmentation fault (core dumped) >> $ s390x-softmmu/qemu-system-s390x -kernel /tmp/test.txt -initrd /tmp/test.txt >> Segmentation fault (core dumped) >> $ echo -n HdrS > /tmp/hdr.txt >> $ sparc64-softmmu/qemu-system-sparc64 -kernel /tmp/hdr.txt -initrd /tmp/hdr.txt >> Segmentation fault (core dumped) >> >> We need a possibility to check the size of the ROM area that we want >> to access, thus let's add a size parameter to the rom_ptr() function >> to avoid these problems. >> >> Signed-off-by: Thomas Huth <thuth@redhat.com> > > Grmpf, this conflicts with "s390/ipl: fix ipl with -no-reboot" in > s390-next... do you want to rebase it? Really? Where is the conflict? I can rebase my patch to your s390-next branch, but the patch looks identical after that rebase. As far as I can see, the patches only change different parts of the ipl.c file. Thomas
On Tue, 26 Jun 2018 10:55:09 +0200 Thomas Huth <thuth@redhat.com> wrote: > On 26.06.2018 10:40, Cornelia Huck wrote: > > On Fri, 15 Jun 2018 11:49:05 +0200 > > Thomas Huth <thuth@redhat.com> wrote: > > > >> The rom_ptr() function allows direct access to the ROM blobs that we > >> load during startup. However, there are currently no checks for the > >> size of the accesses, so it's currently possible to crash QEMU for > >> example with: > >> > >> $ echo "Insane in the mainframe" > /tmp/test.txt > >> $ s390x-softmmu/qemu-system-s390x -kernel /tmp/test.txt -append xyz > >> Segmentation fault (core dumped) > >> $ s390x-softmmu/qemu-system-s390x -kernel /tmp/test.txt -initrd /tmp/test.txt > >> Segmentation fault (core dumped) > >> $ echo -n HdrS > /tmp/hdr.txt > >> $ sparc64-softmmu/qemu-system-sparc64 -kernel /tmp/hdr.txt -initrd /tmp/hdr.txt > >> Segmentation fault (core dumped) > >> > >> We need a possibility to check the size of the ROM area that we want > >> to access, thus let's add a size parameter to the rom_ptr() function > >> to avoid these problems. > >> > >> Signed-off-by: Thomas Huth <thuth@redhat.com> > > > > Grmpf, this conflicts with "s390/ipl: fix ipl with -no-reboot" in > > s390-next... do you want to rebase it? > > Really? Where is the conflict? I can rebase my patch to your s390-next > branch, but the patch looks identical after that rebase. As far as I can > see, the patches only change different parts of the ipl.c file. Yes, but the result does not compile :)
diff --git a/hw/core/loader.c b/hw/core/loader.c index 06bdbca..4a06cce 100644 --- a/hw/core/loader.c +++ b/hw/core/loader.c @@ -191,7 +191,7 @@ void pstrcpy_targphys(const char *name, hwaddr dest, int buf_size, rom_add_blob_fixed(name, source, (nulp - source) + 1, dest); } else { rom_add_blob_fixed(name, source, buf_size, dest); - ptr = rom_ptr(dest + buf_size - 1); + ptr = rom_ptr(dest + buf_size - 1, sizeof(*ptr)); *ptr = 0; } } @@ -1165,7 +1165,7 @@ void rom_reset_order_override(void) fw_cfg_reset_order_override(fw_cfg); } -static Rom *find_rom(hwaddr addr) +static Rom *find_rom(hwaddr addr, size_t size) { Rom *rom; @@ -1179,7 +1179,7 @@ static Rom *find_rom(hwaddr addr) if (rom->addr > addr) { continue; } - if (rom->addr + rom->romsize < addr) { + if (rom->addr + rom->romsize < addr + size) { continue; } return rom; @@ -1249,11 +1249,11 @@ int rom_copy(uint8_t *dest, hwaddr addr, size_t size) return (d + l) - dest; } -void *rom_ptr(hwaddr addr) +void *rom_ptr(hwaddr addr, size_t size) { Rom *rom; - rom = find_rom(addr); + rom = find_rom(addr, size); if (!rom || !rom->data) return NULL; return rom->data + (addr - rom->addr); diff --git a/hw/mips/mips_malta.c b/hw/mips/mips_malta.c index 494f84e..3733e2f 100644 --- a/hw/mips/mips_malta.c +++ b/hw/mips/mips_malta.c @@ -1133,11 +1133,13 @@ void mips_malta_init(MachineState *machine) a neat trick which allows bi-endian firmware. */ #ifndef TARGET_WORDS_BIGENDIAN { - uint32_t *end, *addr = rom_ptr(FLASH_ADDRESS); + uint32_t *end, *addr; + const size_t swapsize = MIN(bios_size, 0x3e0000); + addr = rom_ptr(FLASH_ADDRESS, swapsize); if (!addr) { addr = memory_region_get_ram_ptr(bios); } - end = (void *)addr + MIN(bios_size, 0x3e0000); + end = (void *)addr + swapsize; while (addr < end) { bswap32s(addr); addr++; diff --git a/hw/s390x/ipl.c b/hw/s390x/ipl.c index 04245b5..bcac204 100644 --- a/hw/s390x/ipl.c +++ b/hw/s390x/ipl.c @@ -169,9 +169,12 @@ static void s390_ipl_realize(DeviceState *dev, Error **errp) * loader) and it won't work. For this case we force it to 0x10000, too. */ if (pentry == KERN_IMAGE_START || pentry == 0x800) { + char *parm_area = rom_ptr(KERN_PARM_AREA, strlen(ipl->cmdline) + 1); ipl->start_addr = KERN_IMAGE_START; /* Overwrite parameters in the kernel image, which are "rom" */ - strcpy(rom_ptr(KERN_PARM_AREA), ipl->cmdline); + if (parm_area) { + strcpy(parm_area, ipl->cmdline); + } } else { ipl->start_addr = pentry; } @@ -179,6 +182,7 @@ static void s390_ipl_realize(DeviceState *dev, Error **errp) if (ipl->initrd) { ram_addr_t initrd_offset; int initrd_size; + uint64_t *romptr; initrd_offset = INITRD_START; while (kernel_size + 0x100000 > initrd_offset) { @@ -195,8 +199,11 @@ static void s390_ipl_realize(DeviceState *dev, Error **errp) * we have to overwrite values in the kernel image, * which are "rom" */ - stq_p(rom_ptr(INITRD_PARM_START), initrd_offset); - stq_p(rom_ptr(INITRD_PARM_SIZE), initrd_size); + romptr = rom_ptr(INITRD_PARM_START, 16); + if (romptr) { + stq_p(romptr, initrd_offset); + stq_p(romptr + 1, initrd_size); + } } } /* diff --git a/hw/sparc/sun4m.c b/hw/sparc/sun4m.c index 0ee779f..8eef74b 100644 --- a/hw/sparc/sun4m.c +++ b/hw/sparc/sun4m.c @@ -272,8 +272,8 @@ static unsigned long sun4m_load_kernel(const char *kernel_filename, } if (initrd_size > 0) { for (i = 0; i < 64 * TARGET_PAGE_SIZE; i += TARGET_PAGE_SIZE) { - ptr = rom_ptr(KERNEL_LOAD_ADDR + i); - if (ldl_p(ptr) == 0x48647253) { // HdrS + ptr = rom_ptr(KERNEL_LOAD_ADDR + i, 24); + if (ptr && ldl_p(ptr) == 0x48647253) { /* HdrS */ stl_p(ptr + 16, INITRD_LOAD_ADDR); stl_p(ptr + 20, initrd_size); break; diff --git a/hw/sparc64/sun4u.c b/hw/sparc64/sun4u.c index 1bede85..5b76863 100644 --- a/hw/sparc64/sun4u.c +++ b/hw/sparc64/sun4u.c @@ -186,8 +186,8 @@ static uint64_t sun4u_load_kernel(const char *kernel_filename, } if (*initrd_size > 0) { for (i = 0; i < 64 * TARGET_PAGE_SIZE; i += TARGET_PAGE_SIZE) { - ptr = rom_ptr(*kernel_addr + i); - if (ldl_p(ptr + 8) == 0x48647253) { /* HdrS */ + ptr = rom_ptr(*kernel_addr + i, 32); + if (ptr && ldl_p(ptr + 8) == 0x48647253) { /* HdrS */ stl_p(ptr + 24, *initrd_addr + *kernel_addr); stl_p(ptr + 28, *initrd_size); break; diff --git a/include/hw/loader.h b/include/hw/loader.h index 5ed3fd8..e98b84b 100644 --- a/include/hw/loader.h +++ b/include/hw/loader.h @@ -226,7 +226,7 @@ void rom_set_fw(FWCfgState *f); void rom_set_order_override(int order); void rom_reset_order_override(void); int rom_copy(uint8_t *dest, hwaddr addr, size_t size); -void *rom_ptr(hwaddr addr); +void *rom_ptr(hwaddr addr, size_t size); void hmp_info_roms(Monitor *mon, const QDict *qdict); #define rom_add_file_fixed(_f, _a, _i) \ diff --git a/target/arm/cpu.c b/target/arm/cpu.c index ab047b9..95f92b2 100644 --- a/target/arm/cpu.c +++ b/target/arm/cpu.c @@ -232,7 +232,7 @@ static void arm_cpu_reset(CPUState *s) /* Load the initial SP and PC from offset 0 and 4 in the vector table */ vecbase = env->v7m.vecbase[env->v7m.secure]; - rom = rom_ptr(vecbase); + rom = rom_ptr(vecbase, 8); if (rom) { /* Address zero is covered by ROM which hasn't yet been * copied into physical memory.
The rom_ptr() function allows direct access to the ROM blobs that we load during startup. However, there are currently no checks for the size of the accesses, so it's currently possible to crash QEMU for example with: $ echo "Insane in the mainframe" > /tmp/test.txt $ s390x-softmmu/qemu-system-s390x -kernel /tmp/test.txt -append xyz Segmentation fault (core dumped) $ s390x-softmmu/qemu-system-s390x -kernel /tmp/test.txt -initrd /tmp/test.txt Segmentation fault (core dumped) $ echo -n HdrS > /tmp/hdr.txt $ sparc64-softmmu/qemu-system-sparc64 -kernel /tmp/hdr.txt -initrd /tmp/hdr.txt Segmentation fault (core dumped) We need a possibility to check the size of the ROM area that we want to access, thus let's add a size parameter to the rom_ptr() function to avoid these problems. Signed-off-by: Thomas Huth <thuth@redhat.com> --- v3: Fix the check in find_rom() hw/core/loader.c | 10 +++++----- hw/mips/mips_malta.c | 6 ++++-- hw/s390x/ipl.c | 13 ++++++++++--- hw/sparc/sun4m.c | 4 ++-- hw/sparc64/sun4u.c | 4 ++-- include/hw/loader.h | 2 +- target/arm/cpu.c | 2 +- 7 files changed, 25 insertions(+), 16 deletions(-)