hw/scsi/megasas: Avoid buffer overrun in megasas_handle_scsi()
cover hw/scsi/megasas: Avoid buffer overrun in megasas_handle_scsi()
FWIW megasas is not use by KVM.
Not sure what is the proper fix, but at least we
have a reproducer.
We might improve "scsi/utils" by adding length argument to scsi_cdb_length() and check valid there, but this will take time (large refactor). Add assertions there is too violent.
Philippe Mathieu-Daudé (3):
tests/qtest/fuzz-test: Quit test_lp1878642 once done hw/scsi/megasas: Assert cdb_len is valid in megasas_handle_scsi() hw/scsi/megasas: Have incorrect cdb return MFI_STAT_ABORT_NOT_POSSIBLE
hw/scsi/megasas.c | 6 ++ tests/qtest/fuzz-test.c | 197 ++++++++++++++++++++++++++++++++++++++++ 2 files changed, 203 insertions(+)
Signature · 1 line
2.26.2
PATCH 1/3 tests/qtest/fuzz-test: Quit test_lp1878642 once done
Missed in fd250172842 ("qtest: add a reproducer for LP#1878642").
--- tests/qtest/fuzz-test.c | 1 + 1 file changed, 1 insertion(+)
tests/qtest/fuzz-test.c +1 -0
@@ -45,6 +45,7 @@ static void test_lp1878642_pci_bus_get_irq_level_assert(void)qtest_outl (s,0xcf8 ,0x8400f841 );qtest_outl (s,0xcfc ,0xebed205d );qtest_outl (s,0x5d02 ,0xebed205d ); +qtest_quit (s); }int main (int argc,char **argv)
Signature · 1 line
2.26.2
Re: [PATCH 1/3] tests/qtest/fuzz-test: Quit test_lp1878642 once done
Quoted 13 lines · Philippe Mathieu-Daudé
Missed in
fd250172842("qtest: add a reproducer for LP#1878642").Signed-off-by Philippe Mathieu-Daudé <[email protected]>--- tests/qtest/fuzz-test.c | 1 + 1 file changed, 1 insertion(+)tests/qtest/fuzz-test.c +0
-0
tests/qtest/fuzz-test.c static void test_lp1878642_pci_bus_get_irq_level_assert(void) · line 51
Quoted 7 lines
tests/qtest/fuzz-test.c +1
-0@@ -45,6 +45,7 @@ static void test_lp1878642_pci_bus_get_irq_level_assert(void)qtest_outl (s,0xcf8 ,0x8400f841 );qtest_outl (s,0xcfc ,0xebed205d );qtest_outl (s,0x5d02 ,0xebed205d ); +qtest_quit (s); }int main (int argc,char **argv)
PATCH 2/3 hw/scsi/megasas: Assert cdb_len is valid in megasas_handle_scsi()
In reply to Philippe Mathieu-Daudécdb_len can not be zero... (or less than 6) here, else we have a out-of-bound read first in scsi_cdb_length():
71 int scsi_cdb_length(uint8_t *buf)
72 {
73 int cdb_len;
74
75 switch (buf[0] >> 5) {
76 case 0:
77 cdb_len = 6;
78 break;
Then another out-of-bound read when the size returned by scsi_cdb_length() is used.
Add a reproducer which triggers:
$ make check-qtest-x86_64 Running test qtest-x86_64/fuzz-test qemu-system-x86_64: hw/scsi/megasas.c:1679: megasas_handle_scsi: Assertion `cdb_len > 0 && scsi_cdb_length(cdb) >= cdb_len' failed. tests/qtest/libqtest.c:181: kill_qemu() detected QEMU death from signal 6 (Aborted) (core dumped) ERROR qtest-x86_64/fuzz-test - too few tests run (expected 1, got 0)
--- hw/scsi/megasas.c | 1 + tests/qtest/fuzz-test.c | 196 ++++++++++++++++++++++++++++++++++++++++ 2 files changed, 197 insertions(+)
hw/scsi/megasas.c +1 -0
@@ -1676,6 +1676,7 @@ static int megasas_handle_scsi(MegasasState *s, MegasasCmd *cmd, lun_id = cmd->frame->header.lun_id; cdb_len = cmd->frame->header.cdb_len; +assert (cdb_len >0 &&scsi_cdb_length (cdb) >= cdb_len);if (is_logical) {if (target_id >=MFI_MAX_LD || lun_id !=0 ) { trace_megasas_scsi_target_not_present(
tests/qtest/fuzz-test.c +196 -0
208 lines not on this page · raw message
Signature · 1 line
2.26.2
Re: [PATCH 2/3] hw/scsi/megasas: Assert cdb_len is valid in megasas_handle_scsi()
Quoted 23 lines · Philippe Mathieu-Daudé
cdb_lencan not be zero... (or less than 6) here, else we have a out-of-bound read first inscsi_cdb_length():71 int
scsi_cdb_length(uint8_t *buf)
72 {
73 intcdb_len;
74
75 switch (buf[0] >> 5) {
76 case 0:
77cdb_len= 6;
78 break;Then another out-of-bound read when the size returned by
scsi_cdb_length()is used.Add a reproducer which triggers:
$ make check-qtest-x86_64 Running test qtest-x86_64/fuzz-test qemu-system-x86_64: hw/scsi/megasas.c:1679: megasas_handle_scsi: Assertion `cdb_len > 0 && scsi_cdb_length(cdb) >= cdb_len' failed. tests/qtest/libqtest.c:181: kill_qemu() detected QEMU death from signal 6 (Aborted) (core dumped) ERROR qtest-x86_64/fuzz-test - too few tests run (expected 1, got 0)
Quoted 5 lines
Inspired-by Alexander Bulekov <[email protected]>Signed-off-by Philippe Mathieu-Daudé <[email protected]>--- hw/scsi/megasas.c | 1 + tests/qtest/fuzz-test.c | 196 ++++++++++++++++++++++++++++++++++++++++
For the final version, I think you should add the fuzz-test after the fix (patch 3) ... otherwise this breaks bisection later...
Thomas
PATCH 1/1 hw/scsi/megasas: Assert cdb_len is valid in megasas_handle_scsi()
In reply to Philippe Mathieu-DaudéFrom: Philippe Mathieu-Daudé <[email protected]>
cdb_len can not be zero... (or less than 6) here, else we have a out-of-bound read first in scsi_cdb_length():
71 int scsi_cdb_length(uint8_t *buf)
72 {
73 int cdb_len;
74
75 switch (buf[0] >> 5) {
76 case 0:
77 cdb_len = 6;
78 break;
Then another out-of-bound read when the size returned by scsi_cdb_length() is used.
Add a reproducer which triggers:
$ make check-qtest-x86_64 Running test qtest-x86_64/fuzz-test qemu-system-x86_64: hw/scsi/megasas.c:1679: megasas_handle_scsi: Assertion `cdb_len > 0 && scsi_cdb_length(cdb) >= cdb_len' failed. tests/qtest/libqtest.c:181: kill_qemu() detected QEMU death from signal 6 (Aborted) (core dumped) ERROR qtest-x86_64/fuzz-test - too few tests run (expected 1, got 0)
---
Ran it through the minimizer - No more blobs.
hw/scsi/megasas.c | 1 + tests/qtest/fuzz-test.c | 19 +++++++++++++++++++ 2 files changed, 20 insertions(+)
hw/scsi/megasas.c +1 -0
@@ -1676,6 +1676,7 @@ static int megasas_handle_scsi(MegasasState *s, MegasasCmd *cmd, lun_id = cmd->frame->header.lun_id; cdb_len = cmd->frame->header.cdb_len; +assert (cdb_len >0 &&scsi_cdb_length (cdb) >= cdb_len);if (is_logical) {if (target_id >=MFI_MAX_LD || lun_id !=0 ) { trace_megasas_scsi_target_not_present(
tests/qtest/fuzz-test.c +19 -0
@@ -48,6 +48,23 @@ static void test_lp1878642_pci_bus_get_irq_level_assert(void)qtest_quit (s); } +static void test_megasas_cdb_len_zero (void ) +{ +QTestState *s; + + s =qtest_init ("-M pc -nodefaults " +"-device megasas-gen2 -device scsi-cd,drive=null0 " +"-blockdev driver=null-co,read-zeroes=on,node-name=null0" ); + +qtest_outl (s,0xcf8 ,0x80001011 ); +qtest_outb (s,0xcfc ,0xbb ); +qtest_outl (s,0xcf8 ,0x80001002 ); +qtest_outl (s,0xcfc ,0xf3ff2966 ); +qtest_writeb (s,0x4600 ,0x03 ); +qtest_outw (s,0xbb40 ,0x460b ); +qtest_quit (s); +} +int main (int argc,char **argv) {const char *arch =qtest_get_arch ();
@@ -59,6 +76,8 @@ int main(int argc, char **argv) test_lp1878263_megasas_zero_iov_cnt);qtest_add_func ("fuzz/test_lp1878642_pci_bus_get_irq_level_assert" , test_lp1878642_pci_bus_get_irq_level_assert); +qtest_add_func ("fuzz/test_megasas_cdb_len_zero" , + test_megasas_cdb_len_zero); }return g_test_run ();
Signature · 1 line
2.28.0
RFC PATCH 3/3 hw/scsi/megasas: Have incorrect cdb return MFI_STAT_ABORT_NOT_POSSIBLE
In reply to Philippe Mathieu-DaudéAvoid out-of-bound array access with invalid CDB is provided.
---
RFC because I have no clue how hardware works. Maybe returning MFI_STAT_ARRAY_INDEX_INVALID is better? Do we need to call megasas_write_sense()?
hw/scsi/megasas.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-)
hw/scsi/megasas.c +6 -1
@@ -1676,7 +1676,12 @@ static int megasas_handle_scsi(MegasasState *s, MegasasCmd *cmd, lun_id = cmd->frame->header.lun_id; cdb_len = cmd->frame->header.cdb_len; -assert (cdb_len >0 &&scsi_cdb_length (cdb) >= cdb_len); +if (!cdb_len ||scsi_cdb_length (cdb) < cdb_len) { +trace_megasas_scsi_invalid_cdb_len (mfi_frame_desc (frame_cmd), + is_logical, target_id, + lun_id, cdb_len); +return MFI_STAT_ABORT_NOT_POSSIBLE ; + }if (is_logical) {if (target_id >=MFI_MAX_LD || lun_id !=0 ) { trace_megasas_scsi_target_not_present(
Signature · 1 line
2.26.2