Skip to contentSkip to conversation

ide: Fix incorrect handling of some PRDTs and add the corresponding unit-test

PATCH Applied Alexander Popov 3 participants 13 messages Dec 23, 2019 – Jan 23, 2020

lore
Share & tools
lore
#0 lore raw

v3 cover ide: Fix incorrect handling of some PRDTs and add the corresponding unit-test

Fuzzing the Linux kernel with syzkaller allowed to find how to crash qemu using a special SCSI_IOCTL_SEND_COMMAND. It hits the assertion in ide_dma_cb() introduced in the commit a718978ed58a in July 2015.

This patch series fixes incorrect handling of some PRDTs in ide_dma_cb() and improves the ide-test to cover more PRDT cases (including one that causes that particular qemu crash).

Changes from v2 (thanks to Kevin Wolf for the feedback):
- the assertion about prepare_buf() return value is improved;
- the patch order is reversed to keep the tree bisectable;
- the unit-test performance is improved -- now it runs 8 seconds instead of 3 minutes on my laptop.

Alexander Popov (2):
ide: Fix incorrect handling of some PRDTs in ide_dma_cb() tests/ide-test: Create a single unit-test covering more PRDT cases

 hw/ide/core.c    |  30 +++++---
 tests/ide-test.c | 174 ++++++++++++++++++++---------------------------
 2 files changed, 96 insertions(+), 108 deletions(-)
Signature · 1 line
2.23.0
#1 lore raw

PATCH v3 1/2 ide: Fix incorrect handling of some PRDTs in ide_dma_cb()

The commit a718978ed58a from July 2015 introduced the assertion which implies that the size of successful DMA transfers handled in ide_dma_cb() should be multiple of 512 (the size of a sector). But guest systems can initiate DMA transfers that don't fit this requirement.

For fixing that let's check the number of bytes prepared for the transfer by the prepare_buf() handler. The code in ide_dma_cb() must behave according to the Programming Interface for Bus Master IDE Controller (Revision 1.0 5/16/94):
1. If PRDs specified a smaller size than the IDE transfer size, then the Interrupt and Active bits in the Controller status register are not set (Error Condition).
2. If the size of the physical memory regions was equal to the IDE device transfer size, the Interrupt bit in the Controller status register is set to 1, Active bit is set to 0.
3. If PRDs specified a larger size than the IDE transfer size, the Interrupt and Active bits in the Controller status register are both set to 1.

---
 hw/ide/core.c | 30 ++++++++++++++++++++++--------
 1 file changed, 22 insertions(+), 8 deletions(-)
hw/ide/core.c +22 -8
index 754ff4dc34..80000eb766 100644
@@ -849,6 +849,7 @@ static void ide_dma_cb(void *opaque, int ret)
     int64_t sector_num;
     uint64_t offset;
     bool stay_active = false;
+    int32_t prep_size = 0;
 
     if (ret == -EINVAL) {
         ide_dma_error(s);
@@ -863,13 +864,15 @@ static void ide_dma_cb(void *opaque, int ret)
         }
     }
 
-    n = s->io_buffer_size >> 9;
-    if (n > s->nsector) {
-        /* The PRDs were longer than needed for this request. Shorten them so
-         * we don't get a negative remainder. The Active bit must remain set
-         * after the request completes. */
+    if (s->io_buffer_size > s->nsector * 512) {
+        /*
+         * The PRDs were longer than needed for this request.
+         * The Active bit must remain set after the request completes.
+         */
         n = s->nsector;
         stay_active = true;
+    } else {
+        n = s->io_buffer_size >> 9;
     }
 
     sector_num = ide_get_sector(s);
@@ -892,9 +895,20 @@ static void ide_dma_cb(void *opaque, int ret)
     n = s->nsector;
     s->io_buffer_index = 0;
     s->io_buffer_size = n * 512;
-    if (s->bus->dma->ops->prepare_buf(s->bus->dma, s->io_buffer_size) < 512) {
-        /* The PRDs were too short. Reset the Active bit, but don't raise an
-         * interrupt. */
+    prep_size = s->bus->dma->ops->prepare_buf(s->bus->dma, s->io_buffer_size);
+    /* prepare_buf() must succeed and respect the limit */
+    assert(prep_size >= 0 && prep_size <= n * 512);
+
+    /*
+     * Now prep_size stores the number of bytes in the sglist, and
+     * s->io_buffer_size stores the number of bytes described by the PRDs.
+     */
+
+    if (prep_size < n * 512) {
+        /*
+         * The PRDs are too short for this request. Error condition!
+         * Reset the Active bit and don't raise the interrupt.
+         */
         s->status = READY_STAT | SEEK_STAT;
         dma_buf_commit(s, 0);
         goto eot;
Signature · 1 line
2.23.0
Reviewed-by on v3 1/2
#2 lore raw

Re: [PATCH v3 1/2] ide: Fix incorrect handling of some PRDTs in ide_dma_cb()

Am 23.12.2019 um 18:51 hat Alexander Popov geschrieben:

Quoted 19 lines

The commit a718978ed58a from July 2015 introduced the assertion which implies that the size of successful DMA transfers handled in ide_dma_cb() should be multiple of 512 (the size of a sector). But guest systems can initiate DMA transfers that don't fit this requirement.

For fixing that let's check the number of bytes prepared for the transfer by the prepare_buf() handler. The code in ide_dma_cb() must behave according to the Programming Interface for Bus Master IDE Controller (Revision 1.0 5/16/94):
1. If PRDs specified a smaller size than the IDE transfer size, then the Interrupt and Active bits in the Controller status register are not set (Error Condition).
2. If the size of the physical memory regions was equal to the IDE device transfer size, the Interrupt bit in the Controller status register is set to 1, Active bit is set to 0.
3. If PRDs specified a larger size than the IDE transfer size, the Interrupt and Active bits in the Controller status register are both set to 1.

Quoted 1 line
#3 lore raw

PATCH v3 2/2 tests/ide-test: Create a single unit-test covering more PRDT cases

In reply to Alexander Popov

Fuzzing the Linux kernel with syzkaller allowed to find how to crash qemu using a special SCSI_IOCTL_SEND_COMMAND. It hits the assertion in ide_dma_cb() introduced in the commit a718978ed58a in July 2015. Currently this bug is not reproduced by the unit tests.

Let's improve the ide-test to cover more PRDT cases including one that causes this particular qemu crash.

The test is developed according to the Programming Interface for Bus Master IDE Controller (Revision 1.0 5/16/94).

---
 tests/ide-test.c | 174 ++++++++++++++++++++---------------------------
 1 file changed, 74 insertions(+), 100 deletions(-)
tests/ide-test.c +74 -100
index 0277e7d5a9..5cfd97f915 100644
@@ -445,104 +445,81 @@ static void test_bmdma_trim(void)
     test_bmdma_teardown(qts);
 }
 
-static void test_bmdma_short_prdt(void)
-{
-    QTestState *qts;
-    QPCIDevice *dev;
-    QPCIBar bmdma_bar, ide_bar;
-    uint8_t status;
-
-    PrdtEntry prdt[] = {
-        {
-            .addr = 0,
-            .size = cpu_to_le32(0x10 | PRDT_EOT),
-        },
-    };
-
-    qts = test_bmdma_setup();
-
-    dev = get_pci_device(qts, &bmdma_bar, &ide_bar);
-
-    /* Normal request */
-    status = send_dma_request(qts, CMD_READ_DMA, 0, 1,
-                              prdt, ARRAY_SIZE(prdt), NULL);
-    g_assert_cmphex(status, ==, 0);
-    assert_bit_clear(qpci_io_readb(dev, ide_bar, reg_status), DF | ERR);
-
-    /* Abort the request before it completes */
-    status = send_dma_request(qts, CMD_READ_DMA | CMDF_ABORT, 0, 1,
-                              prdt, ARRAY_SIZE(prdt), NULL);
-    g_assert_cmphex(status, ==, 0);
-    assert_bit_clear(qpci_io_readb(dev, ide_bar, reg_status), DF | ERR);
-    free_pci_device(dev);
-    test_bmdma_teardown(qts);
-}
-
-static void test_bmdma_one_sector_short_prdt(void)
-{
-    QTestState *qts;
-    QPCIDevice *dev;
-    QPCIBar bmdma_bar, ide_bar;
-    uint8_t status;
-
-    /* Read 2 sectors but only give 1 sector in PRDT */
-    PrdtEntry prdt[] = {
-        {
-            .addr = 0,
-            .size = cpu_to_le32(0x200 | PRDT_EOT),
-        },
-    };
-
-    qts = test_bmdma_setup();
-
-    dev = get_pci_device(qts, &bmdma_bar, &ide_bar);
-
-    /* Normal request */
-    status = send_dma_request(qts, CMD_READ_DMA, 0, 2,
-                              prdt, ARRAY_SIZE(prdt), NULL);
-    g_assert_cmphex(status, ==, 0);
-    assert_bit_clear(qpci_io_readb(dev, ide_bar, reg_status), DF | ERR);
-
-    /* Abort the request before it completes */
-    status = send_dma_request(qts, CMD_READ_DMA | CMDF_ABORT, 0, 2,
-                              prdt, ARRAY_SIZE(prdt), NULL);
-    g_assert_cmphex(status, ==, 0);
-    assert_bit_clear(qpci_io_readb(dev, ide_bar, reg_status), DF | ERR);
-    free_pci_device(dev);
-    test_bmdma_teardown(qts);
-}
-
-static void test_bmdma_long_prdt(void)
+/*
+ * This test is developed according to the Programming Interface for
+ * Bus Master IDE Controller (Revision 1.0 5/16/94)
+ */
+static void test_bmdma_various_prdts(void)
 {
-    QTestState *qts;
-    QPCIDevice *dev;
-    QPCIBar bmdma_bar, ide_bar;
-    uint8_t status;
-
-    PrdtEntry prdt[] = {
-        {
-            .addr = 0,
-            .size = cpu_to_le32(0x1000 | PRDT_EOT),
-        },
-    };
-
-    qts = test_bmdma_setup();
-
-    dev = get_pci_device(qts, &bmdma_bar, &ide_bar);
-
-    /* Normal request */
-    status = send_dma_request(qts, CMD_READ_DMA, 0, 1,
-                              prdt, ARRAY_SIZE(prdt), NULL);
-    g_assert_cmphex(status, ==, BM_STS_ACTIVE | BM_STS_INTR);
-    assert_bit_clear(qpci_io_readb(dev, ide_bar, reg_status), DF | ERR);
+    int sectors = 0;
+    uint32_t size = 0;
+
+    for (sectors = 1; sectors <= 256; sectors *= 2) {
+        QTestState *qts = NULL;
+        QPCIDevice *dev = NULL;
+        QPCIBar bmdma_bar, ide_bar;
+
+        qts = test_bmdma_setup();
+        dev = get_pci_device(qts, &bmdma_bar, &ide_bar);
+
+        for (size = 0; size < 65536; size += 256) {
+            uint32_t req_size = sectors * 512;
+            uint32_t prd_size = size & 0xfffe; /* bit 0 is always set to 0 */
+            uint8_t ret = 0;
+            uint8_t req_status = 0;
+            uint8_t abort_req_status = 0;
+            PrdtEntry prdt[] = {
+                {
+                    .addr = 0,
+                    .size = cpu_to_le32(size | PRDT_EOT),
+                },
+            };
+
+            /* A value of zero in PRD size indicates 64K */
+            if (prd_size == 0) {
+                prd_size = 65536;
+            }
+
+            /*
+             * 1. If PRDs specified a smaller size than the IDE transfer
+             * size, then the Interrupt and Active bits in the Controller
+             * status register are not set (Error Condition).
+             *
+             * 2. If the size of the physical memory regions was equal to
+             * the IDE device transfer size, the Interrupt bit in the
+             * Controller status register is set to 1, Active bit is set to 0.
+             *
+             * 3. If PRDs specified a larger size than the IDE transfer size,
+             * the Interrupt and Active bits in the Controller status register
+             * are both set to 1.
+             */
+            if (prd_size < req_size) {
+                req_status = 0;
+                abort_req_status = 0;
+            } else if (prd_size == req_size) {
+                req_status = BM_STS_INTR;
+                abort_req_status = BM_STS_INTR;
+            } else {
+                req_status = BM_STS_ACTIVE | BM_STS_INTR;
+                abort_req_status = BM_STS_INTR;
+            }
+
+            /* Test the request */
+            ret = send_dma_request(qts, CMD_READ_DMA, 0, sectors,
+                                   prdt, ARRAY_SIZE(prdt), NULL);
+            g_assert_cmphex(ret, ==, req_status);
+            assert_bit_clear(qpci_io_readb(dev, ide_bar, reg_status), DF | ERR);
+
+            /* Now test aborting the same request */
+            ret = send_dma_request(qts, CMD_READ_DMA | CMDF_ABORT, 0,
+                                   sectors, prdt, ARRAY_SIZE(prdt), NULL);
+            g_assert_cmphex(ret, ==, abort_req_status);
+            assert_bit_clear(qpci_io_readb(dev, ide_bar, reg_status), DF | ERR);
+        }
 
-    /* Abort the request before it completes */
-    status = send_dma_request(qts, CMD_READ_DMA | CMDF_ABORT, 0, 1,
-                              prdt, ARRAY_SIZE(prdt), NULL);
-    g_assert_cmphex(status, ==, BM_STS_INTR);
-    assert_bit_clear(qpci_io_readb(dev, ide_bar, reg_status), DF | ERR);
-    free_pci_device(dev);
-    test_bmdma_teardown(qts);
+        free_pci_device(dev);
+        test_bmdma_teardown(qts);
+    }
 }
 
 static void test_bmdma_no_busmaster(void)
@@ -1066,10 +1043,7 @@ int main(int argc, char **argv)
 
     qtest_add_func("/ide/bmdma/simple_rw", test_bmdma_simple_rw);
     qtest_add_func("/ide/bmdma/trim", test_bmdma_trim);
-    qtest_add_func("/ide/bmdma/short_prdt", test_bmdma_short_prdt);
-    qtest_add_func("/ide/bmdma/one_sector_short_prdt",
-                   test_bmdma_one_sector_short_prdt);
-    qtest_add_func("/ide/bmdma/long_prdt", test_bmdma_long_prdt);
+    qtest_add_func("/ide/bmdma/various_prdts", test_bmdma_various_prdts);
     qtest_add_func("/ide/bmdma/no_busmaster", test_bmdma_no_busmaster);
 
     qtest_add_func("/ide/flush", test_flush);
Signature · 1 line
2.23.0
#4 lore raw

Re: [PATCH v3 2/2] tests/ide-test: Create a single unit-test covering more PRDT cases

Am 23.12.2019 um 18:51 hat Alexander Popov geschrieben:

Quoted 5 lines

Fuzzing the Linux kernel with syzkaller allowed to find how to crash qemu using a special SCSI_IOCTL_SEND_COMMAND. It hits the assertion in ide_dma_cb() introduced in the commit a718978ed58a in July 2015. Currently this bug is not reproduced by the unit tests.

Quoted 7 lines

Let's improve the ide-test to cover more PRDT cases including one that causes this particular qemu crash.

The test is developed according to the Programming Interface for Bus Master IDE Controller (Revision 1.0 5/16/94).

The time this test takes is much better now (~5s for me).

Quoted 27 lines
tests/ide-test.c +5 -21
+/*
+ * This test is developed according to the Programming Interface for
+ * Bus Master IDE Controller (Revision 1.0 5/16/94)
+ */
+static void test_bmdma_various_prdts(void)
 {
-    QTestState *qts;
-    QPCIDevice *dev;
-    QPCIBar bmdma_bar, ide_bar;
-    uint8_t status;
-
-    PrdtEntry prdt[] = {
-        {
-            .addr = 0,
-            .size = cpu_to_le32(0x1000 | PRDT_EOT),
-        },
-    };
-
-    qts = test_bmdma_setup();
-
-    dev = get_pci_device(qts, &bmdma_bar, &ide_bar);
-
-    /* Normal request */
-    status = send_dma_request(qts, CMD_READ_DMA, 0, 1,
-                              prdt, ARRAY_SIZE(prdt), NULL);
-    g_assert_cmphex(status, ==, BM_STS_ACTIVE | BM_STS_INTR);
-    assert_bit_clear(qpci_io_readb(dev, ide_bar, reg_status), DF | ERR);

tests/ide-test.c static void test_bmdma_trim(void) · line 463

Quoted 10 lines
tests/ide-test.c +10 -0
+    int sectors = 0;
+    uint32_t size = 0;
+
+    for (sectors = 1; sectors <= 256; sectors *= 2) {
+        QTestState *qts = NULL;
+        QPCIDevice *dev = NULL;
+        QPCIBar bmdma_bar, ide_bar;
+
+        qts = test_bmdma_setup();
+        dev = get_pci_device(qts, &bmdma_bar, &ide_bar);

I'm wondering why the initialisation has to be inside the outer for loop. I expected that moving it outside would further improve the speed. But sure enough, doing that makes the test fail.

Did you have a look why this happens? I suppose we might be running out of some resources in the qtest framework becasue each send_dma_request() calls get_pci_device() again?

5 seconds isn't that bad, so this shouldn't block this series, but it's still by far the slowest test in ide-test, so any improvement certainly wouldn't hurt.

tests/ide-test.c static void test_bmdma_trim(void) · line 469

Quoted 5 lines
tests/ide-test.c +5 -0
+        for (size = 0; size < 65536; size += 256) {
+            uint32_t req_size = sectors * 512;
+            uint32_t prd_size = size & 0xfffe; /* bit 0 is always set to 0 */
+            uint8_t ret = 0;
+            uint8_t req_status = 0;

If you end up sending another version for some reason, I would also consider renaming req_status, because reg_status already exists, which looks almost the same. This confused me for a moment when reading the code below.

Kevin

#5 lore raw

Re: [PATCH v3 2/2] tests/ide-test: Create a single unit-test covering more PRDT cases

Quoted 49 lines · Kevin Wolf

Am 23.12.2019 um 18:51 hat Alexander Popov geschrieben:

Quoted 5 lines

Fuzzing the Linux kernel with syzkaller allowed to find how to crash qemu using a special SCSI_IOCTL_SEND_COMMAND. It hits the assertion in ide_dma_cb() introduced in the commit a718978ed58a in July 2015. Currently this bug is not reproduced by the unit tests.

Quoted 7 lines

Let's improve the ide-test to cover more PRDT cases including one that causes this particular qemu crash.

The test is developed according to the Programming Interface for Bus Master IDE Controller (Revision 1.0 5/16/94).

The time this test takes is much better now (~5s for me).

Quoted 33 lines
tests/ide-test.c +11 -21
+/*
+ * This test is developed according to the Programming Interface for
+ * Bus Master IDE Controller (Revision 1.0 5/16/94)
+ */
+static void test_bmdma_various_prdts(void)
 {
-    QTestState *qts;
-    QPCIDevice *dev;
-    QPCIBar bmdma_bar, ide_bar;
-    uint8_t status;
-
-    PrdtEntry prdt[] = {
-        {
-            .addr = 0,
-            .size = cpu_to_le32(0x1000 | PRDT_EOT),
-        },
-    };
-
-    qts = test_bmdma_setup();
-
-    dev = get_pci_device(qts, &bmdma_bar, &ide_bar);
-
-    /* Normal request */
-    status = send_dma_request(qts, CMD_READ_DMA, 0, 1,
-                              prdt, ARRAY_SIZE(prdt), NULL);
-    g_assert_cmphex(status, ==, BM_STS_ACTIVE | BM_STS_INTR);
-    assert_bit_clear(qpci_io_readb(dev, ide_bar, reg_status), DF | ERR);
+    int sectors = 0;
+    uint32_t size = 0;
+
+    for (sectors = 1; sectors <= 256; sectors *= 2) {
+        QTestState *qts = NULL;
+        QPCIDevice *dev = NULL;
Quoted 8 lines

tests/ide-test.c static void test_bmdma_trim(void) · line 463

Quoted 4 lines
tests/ide-test.c +4 -0
+        QPCIBar bmdma_bar, ide_bar;
+
+        qts = test_bmdma_setup();
+        dev = get_pci_device(qts, &bmdma_bar, &ide_bar);

I'm wondering why the initialisation has to be inside the outer for loop. I expected that moving it outside would further improve the speed. But sure enough, doing that makes the test fail.

Yes, that's why I came to the current solution.

Quoted 3 lines

Did you have a look why this happens? I suppose we might be running out of some resources in the qtest framework becasue each send_dma_request() calls get_pci_device() again?

I've spent some time on investigating, but didn't succeed.

1. After several hundreds of send_dma_request() calls the following assertion in that function fails:

    assert_bit_clear(qpci_io_readb(dev, ide_bar, reg_status), BSY | DRQ);

2. If I comment out this assertion, the test system proceeds but eventually stalls.

3. I tried to send the CMD_FLUSH_CACHE command to the device, it didn't help.

4. That behavior is not influenced by ide_dma_cb() code that I changed.

I guess it would be better if that effect is examined by somebody with more knowledge about DMA and qtest.

Quoted 3 lines

5 seconds isn't that bad, so this shouldn't block this series, but it's still by far the slowest test in ide-test, so any improvement certainly wouldn't hurt.

Thanks for not making that mandatory. It would take me much more time.

Quoted 10 lines

tests/ide-test.c static void test_bmdma_trim(void) · line 469

Quoted 5 lines
tests/ide-test.c +5 -0
+        for (size = 0; size < 65536; size += 256) {
+            uint32_t req_size = sectors * 512;
+            uint32_t prd_size = size & 0xfffe; /* bit 0 is always set to 0 */
+            uint8_t ret = 0;
+            uint8_t req_status = 0;

If you end up sending another version for some reason, I would also consider renaming req_status, because reg_status already exists, which looks almost the same. This confused me for a moment when reading the code below.

Heh! Ok, let's wait for more reviews.

Best regards,
Alexander

#6 lore raw

Re: [PATCH v3 2/2] tests/ide-test: Create a single unit-test covering more PRDT cases

Am 07.01.2020 um 23:39 hat Alexander Popov geschrieben:

Quoted 16 lines
Quoted 3 lines

Did you have a look why this happens? I suppose we might be running out of some resources in the qtest framework becasue each send_dma_request() calls get_pci_device() again?

I've spent some time on investigating, but didn't succeed.

1. After several hundreds of send_dma_request() calls the following assertion in that function fails:

    assert_bit_clear(qpci_io_readb(dev, ide_bar, reg_status), BSY | DRQ);

2. If I comment out this assertion, the test system proceeds but eventually stalls.

3. I tried to send the CMD_FLUSH_CACHE command to the device, it didn't help.

4. That behavior is not influenced by ide_dma_cb() code that I changed.

Quoted 8 lines

I guess it would be better if that effect is examined by somebody with more knowledge about DMA and qtest.

Quoted 3 lines

5 seconds isn't that bad, so this shouldn't block this series, but it's still by far the slowest test in ide-test, so any improvement certainly wouldn't hurt.

Thanks for not making that mandatory. It would take me much more time.

Ok, don't bother then.

I seem to remember that I ran into something similar some time ago and found out that it was related to some integer overflow, I think during the PCI BAR mapping. This might be the same. Maybe I'll have another look later.

Kevin

#7 lore raw
In reply to Alexander Popov
Quoted 14 lines · Alexander Popov

Fuzzing the Linux kernel with syzkaller allowed to find how to crash qemu using a special SCSI_IOCTL_SEND_COMMAND. It hits the assertion in ide_dma_cb() introduced in the commit a718978ed58a in July 2015.

This patch series fixes incorrect handling of some PRDTs in ide_dma_cb() and improves the ide-test to cover more PRDT cases (including one that causes that particular qemu crash).

Changes from v2 (thanks to Kevin Wolf for the feedback):
- the assertion about prepare_buf() return value is improved;
- the patch order is reversed to keep the tree bisectable;
- the unit-test performance is improved -- now it runs 8 seconds instead of 3 minutes on my laptop.

Quoted 7 lines

Alexander Popov (2):
ide: Fix incorrect handling of some PRDTs in ide_dma_cb() tests/ide-test: Create a single unit-test covering more PRDT cases

 hw/ide/core.c    |  30 +++++---
 tests/ide-test.c | 174 ++++++++++++++++++++---------------------------
 2 files changed, 96 insertions(+), 108 deletions(-)

Hello!

Just a friendly ping.

Could I have the feedback for this patch series?

Kevin, do you like the changes?

Best regards,
Alexander

#8 lore raw
In reply to Alexander Popov
Quoted 14 lines · Alexander Popov

Fuzzing the Linux kernel with syzkaller allowed to find how to crash qemu using a special SCSI_IOCTL_SEND_COMMAND. It hits the assertion in ide_dma_cb() introduced in the commit a718978ed58a in July 2015.

This patch series fixes incorrect handling of some PRDTs in ide_dma_cb() and improves the ide-test to cover more PRDT cases (including one that causes that particular qemu crash).

Changes from v2 (thanks to Kevin Wolf for the feedback):
- the assertion about prepare_buf() return value is improved;
- the patch order is reversed to keep the tree bisectable;
- the unit-test performance is improved -- now it runs 8 seconds instead of 3 minutes on my laptop.

Quoted 7 lines

Alexander Popov (2):
ide: Fix incorrect handling of some PRDTs in ide_dma_cb() tests/ide-test: Create a single unit-test covering more PRDT cases

 hw/ide/core.c    |  30 +++++---
 tests/ide-test.c | 174 ++++++++++++++++++++---------------------------
 2 files changed, 96 insertions(+), 108 deletions(-)

Hello!

Pinging again about this fix and unit-test...

It's ready. Kevin Wolf has reviewed this (thanks a lot!).

What is next?

Best regards,
Alexander

#9 lore raw

Am 22.01.2020 um 12:53 hat Alexander Popov geschrieben:

Quoted 23 lines · Alexander Popov
Quoted 14 lines

Fuzzing the Linux kernel with syzkaller allowed to find how to crash qemu using a special SCSI_IOCTL_SEND_COMMAND. It hits the assertion in ide_dma_cb() introduced in the commit a718978ed58a in July 2015.

This patch series fixes incorrect handling of some PRDTs in ide_dma_cb() and improves the ide-test to cover more PRDT cases (including one that causes that particular qemu crash).

Changes from v2 (thanks to Kevin Wolf for the feedback):
- the assertion about prepare_buf() return value is improved;
- the patch order is reversed to keep the tree bisectable;
- the unit-test performance is improved -- now it runs 8 seconds instead of 3 minutes on my laptop.

Quoted 7 lines

Alexander Popov (2):
ide: Fix incorrect handling of some PRDTs in ide_dma_cb() tests/ide-test: Create a single unit-test covering more PRDT cases

 hw/ide/core.c    |  30 +++++---
 tests/ide-test.c | 174 ++++++++++++++++++++---------------------------
 2 files changed, 96 insertions(+), 108 deletions(-)
Quoted 7 lines

Hello!

Pinging again about this fix and unit-test...

It's ready. Kevin Wolf has reviewed this (thanks a lot!).

What is next?

I asked John about it just yesterday (if he will merge it or if he would prefer me to take it through my tree) and he promised to take a look very soon.

Kevin

#10 lore raw
Quoted 30 lines · Kevin Wolf

Am 22.01.2020 um 12:53 hat Alexander Popov geschrieben:

Quoted 29 lines · Alexander Popov
Quoted 14 lines

Fuzzing the Linux kernel with syzkaller allowed to find how to crash qemu using a special SCSI_IOCTL_SEND_COMMAND. It hits the assertion in ide_dma_cb() introduced in the commit a718978ed58a in July 2015.

This patch series fixes incorrect handling of some PRDTs in ide_dma_cb() and improves the ide-test to cover more PRDT cases (including one that causes that particular qemu crash).

Changes from v2 (thanks to Kevin Wolf for the feedback):
- the assertion about prepare_buf() return value is improved;
- the patch order is reversed to keep the tree bisectable;
- the unit-test performance is improved -- now it runs 8 seconds instead of 3 minutes on my laptop.

Quoted 7 lines

Alexander Popov (2):
ide: Fix incorrect handling of some PRDTs in ide_dma_cb() tests/ide-test: Create a single unit-test covering more PRDT cases

 hw/ide/core.c    |  30 +++++---
 tests/ide-test.c | 174 ++++++++++++++++++++---------------------------
 2 files changed, 96 insertions(+), 108 deletions(-)

Hello!

Pinging again about this fix and unit-test...

It's ready. Kevin Wolf has reviewed this (thanks a lot!).

Quoted 8 lines
Quoted 1 line

What is next?

I asked John about it just yesterday (if he will merge it or if he would prefer me to take it through my tree) and he promised to take a look very soon.

Kevin

Going to merge it today.

--js

Applied
#11 lore raw
In reply to Alexander Popov
Quoted 14 lines · Alexander Popov

Fuzzing the Linux kernel with syzkaller allowed to find how to crash qemu using a special SCSI_IOCTL_SEND_COMMAND. It hits the assertion in ide_dma_cb() introduced in the commit a718978ed58a in July 2015.

This patch series fixes incorrect handling of some PRDTs in ide_dma_cb() and improves the ide-test to cover more PRDT cases (including one that causes that particular qemu crash).

Changes from v2 (thanks to Kevin Wolf for the feedback):
- the assertion about prepare_buf() return value is improved;
- the patch order is reversed to keep the tree bisectable;
- the unit-test performance is improved -- now it runs 8 seconds instead of 3 minutes on my laptop.

Quoted 8 lines

Alexander Popov (2):
ide: Fix incorrect handling of some PRDTs in ide_dma_cb() tests/ide-test: Create a single unit-test covering more PRDT cases

 hw/ide/core.c    |  30 +++++---
 tests/ide-test.c | 174 ++++++++++++++++++++---------------------------
 2 files changed, 96 insertions(+), 108 deletions(-)

Thanks, applied to my IDE tree:

https://github.com/jnsnow/qemu/commits/ide
https://github.com/jnsnow/qemu.git

--js

#12 lore raw
Quoted 24 lines · John Snow
Quoted 14 lines · Alexander Popov

Fuzzing the Linux kernel with syzkaller allowed to find how to crash qemu using a special SCSI_IOCTL_SEND_COMMAND. It hits the assertion in ide_dma_cb() introduced in the commit a718978ed58a in July 2015.

This patch series fixes incorrect handling of some PRDTs in ide_dma_cb() and improves the ide-test to cover more PRDT cases (including one that causes that particular qemu crash).

Changes from v2 (thanks to Kevin Wolf for the feedback):
- the assertion about prepare_buf() return value is improved;
- the patch order is reversed to keep the tree bisectable;
- the unit-test performance is improved -- now it runs 8 seconds instead of 3 minutes on my laptop.

Quoted 8 lines

Alexander Popov (2):
ide: Fix incorrect handling of some PRDTs in ide_dma_cb() tests/ide-test: Create a single unit-test covering more PRDT cases

 hw/ide/core.c    |  30 +++++---
 tests/ide-test.c | 174 ++++++++++++++++++++---------------------------
 2 files changed, 96 insertions(+), 108 deletions(-)
Quoted 4 lines

Thanks, applied to my IDE tree:

https://github.com/jnsnow/qemu/commits/ide
https://github.com/jnsnow/qemu.git

Happy end!
Thanks a lot!

Best regards,
Alexander