Skip to contentSkip to conversation

scsi: target: fix iSCSI ISID use-after-free in REGISTER AND MOVE

PATCH In review Bryam Vargas 4 participants 4 messages Jun 10

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH v3 1/1 scsi: target: fix iSCSI ISID use-after-free in REGISTER AND MOVE

core_scsi3_emulate_pro_register_and_move() maps the PERSISTENT RESERVE OUT parameter list with transport_kmap_data_sg() and parses the destination TransportID with target_parse_pr_out_transport_id(). For an iSCSI TransportID (FORMAT CODE 01b), iscsi_parse_pr_out_transport_id() returns the ISID in iport_ptr as a raw pointer into that mapped buffer.

The function then unmaps the buffer with transport_kunmap_data_sg() before dereferencing iport_ptr in strcmp(), __core_scsi3_locate_pr_reg() and core_scsi3_alloc_registration(). When the parameter list spans more than one page (PARAMETER LIST LENGTH > 4096), transport_kmap_data_sg() uses vmap() and transport_kunmap_data_sg() does vunmap(), so the kernel virtual address backing iport_ptr is torn down and every subsequent dereference is a use-after-free read of the unmapped region.

Keep the parameter list mapped until iport_ptr is no longer needed: drop the early transport_kunmap_data_sg() and unmap once on the success path, right before returning. The error paths already unmap through the existing "if (buf) transport_kunmap_data_sg(cmd)" at the out: label, which now runs on every post-map error exit because buf is no longer cleared early. Only reads of the mapping happen while spinlocks are held; the map and unmap calls remain outside any lock. The sibling caller core_scsi3_decode_spec_i_port() already uses the buffer before unmapping it and is left unchanged.

Fixes 4949314c7283 ("target: Allow control CDBs with data > 1 page")

---
v3 (review of v2 by John Garry and James Bottomley):
- Drop the parser-ownership approach. Rather than copy the ISID into an allocation that both callers must kfree() (v2), keep the PR-OUT parameter list mapped across the iport_ptr dereferences and move the single unmap to the success path. No allocation, no kfree, and target_core_fabric_lib.c / core_scsi3_decode_spec_i_port() are unchanged. This is the form John Garry asked for ("keep the mapping in place for longer, until the out: label") and removes the multiple kfree() paths v2 added.
- Re: holding the mapping while spinlocks are held (raised on v2): only reads of the mapping occur under dev_reservation_lock; the kmap/kunmap calls are all outside any lock, so there is no sleep-in-atomic concern.

v2: https://lore.kernel.org/linux-scsi/[email protected]/
v1: https://lore.kernel.org/linux-scsi/[email protected]/

Class / impact: CWE-416 use-after-free (use-after-vunmap) in the LIO SCSI target, reachable by an authenticated iSCSI initiator that is a current Persistent Reservation registrant on the LUN. It sends PERSISTENT RESERVE OUT / REGISTER AND MOVE with an iSCSI (FORMAT CODE 01b) TransportID and a PARAMETER LIST LENGTH > 4096 so the parameter list spans >1 page and is mapped with vmap(). After transport_kunmap_data_sg() vunmap()s that region, the retained iport_ptr is dereferenced -> kernel read of an unmapped vmalloc address (oops / DoS). Primarily a remotely reachable authenticated denial of service. Present in all maintained trees since 4949314c7283 (v3.3, 2012), which introduced the multi-page vmap() path. Verified at mainline v7.1-rc6 and stable v6.12.92.

Reproducer (authenticated iSCSI initiator, current PR reservation holder):
1. PERSISTENT RESERVE OUT / REGISTER a key from the iSCSI nexus.
2. PERSISTENT RESERVE OUT / REGISTER AND MOVE, FORMAT CODE 01b TransportID

     (IQN + ",i,0x" + 12-char ISID), RELATIVE TARGET PORT IDENTIFIER of an
     existing target port, with PARAMETER LIST LENGTH = 8192 (two pages ->
     vmap()/vunmap()), the inner ADDITIONAL LENGTH set so tid_len + 24 ==
     data_length, the remainder zero padding.

A/B verification (CONFIG_KASAN_VMALLOC=y, kasan.fault=report, x86-64, 6.12.90; same kernel for every arm; 64-bit and 32-bit initiator):
- Without this patch (8192-byte, two-page request):

      BUG: KASAN: vmalloc-out-of-bounds in strcmp+0xa7/0xb0
        strcmp
        core_scsi3_emulate_pro_register_and_move [target_core]
        ? remove_vm_area
        target_scsi3_emulate_pr_out [target_core]
        __target_execute_cmd / iscsit_execute_cmd / iscsi_target_rx_thread
      followed by "unable to handle page fault" (PTE 0); the command never
      completes (the iSCSI rx kthread dies).
  - Control (128-byte, single-page request): no report (kunmap is a no-op
    on 64-bit !HIGHMEM, no vunmap).
  - With this patch (same 8192-byte request, 64-bit and 32-bit initiator):
    no report, the command completes.
 drivers/target/target_core_pr.c | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)
drivers/target/target_core_pr.c +5 -3
index 11790f2c5d80..7c5bb7d67947 100644
@@ -3293,9 +3293,6 @@ core_scsi3_emulate_pro_register_and_move(struct se_cmd *cmd, u64 res_key,
 		goto out;
 	}
 
-	transport_kunmap_data_sg(cmd);
-	buf = NULL;
-
 	pr_debug("SPC-3 PR [%s] Extracted initiator %s identifier: %s"
 		" %s\n", dest_tf_ops->fabric_name, (iport_ptr != NULL) ?
 		"port" : "device", initiator_str, (iport_ptr != NULL) ?
@@ -3532,6 +3529,11 @@ core_scsi3_emulate_pro_register_and_move(struct se_cmd *cmd, u64 res_key,
 	core_scsi3_update_and_write_aptpl(cmd->se_dev, aptpl);
 
 	core_scsi3_put_pr_reg(dest_pr_reg);
+	/*
+	 * iport_ptr aliases the PR-OUT parameter list mapped above, so the
+	 * buffer is unmapped only here on success (and at out: on error).
+	 */
+	transport_kunmap_data_sg(cmd);
 	return 0;
 out:
 	if (buf)
Signature · 1 line
2.43.0
#1 lore raw
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:

Pre-existing issues:
- [Critical] Missing length validation on the TransportID payload leads to out-of-bounds reads and potential kernel oops via strstr() and get_unaligned_be64(). --

commit f95f1e5f1cdc7f67d3f1cb876ff0263d03032d49
Author: Bryam Vargas <[email protected]>

scsi: target: fix iSCSI ISID use-after-free in REGISTER AND MOVE

This commit fixes a use-after-free read of the iscsi isid pointer in core_scsi3_emulate_pro_register_and_move(). The early unmap of the buffer is dropped, keeping the parameter list mapped until the pointer is no longer needed.

drivers/target/target_core_pr.c core_scsi3_emulate_pro_register_and_move(struct se_cmd *cmd, u64 res_key, · line 3298

Quoted 11 lines
drivers/target/target_core_pr.c +0 -3
index 11790f2c5d80f..7c5bb7d679477 100644
@@ -3293,9 +3293,6 @@ core_scsi3_emulate_pro_register_and_move(struct se_cmd *cmd, u64 res_key,
 		goto out;
 	}
 
-	transport_kunmap_data_sg(cmd);
-	buf = NULL;
-

[Severity: Critical]
This is a pre-existing issue, but does this code properly validate the payload length before parsing the buffer?

Earlier in core_scsi3_emulate_pro_register_and_move(), proto_ident is read from buf[24] without verifying that tid_len > 0. If a command is sent with tid_len = 0 and data_length = 24, would reading buf[24] result in an out-of-bounds read?

Furthermore, iscsi_parse_pr_out_transport_id() calls:

    p = strstr(&buf[4], ",i,0x");

If a payload length is exactly aligned to a multiple of the page size with no null terminator, could strstr() scan past the vmap buffer and hit a guard page, causing a kernel oops?

Finally, if the parsed isid string is found but terminates quickly near a page boundary (for example, ",i,0x1\0"), __core_scsi3_do_alloc_registration() executes:

    pr_reg->pr_reg_bin_isid = get_unaligned_be64(isid);

Since get_unaligned_be64() unconditionally reads 8 bytes from the potentially short ascii string, could this read cross into a guard page and cause an oops?

Quoted 3 lines
drivers/target/target_core_pr.c +0 -0
 	pr_debug("SPC-3 PR [%s] Extracted initiator %s identifier: %s"
 		" %s\n", dest_tf_ops->fabric_name, (iport_ptr != NULL) ?
 		"port" : "device", initiator_str, (iport_ptr != NULL) ?
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1
Reviewed-by on v3 1/1
#2 lore raw
In reply to Bryam Vargas
Quoted 24 lines · Bryam Vargas

core_scsi3_emulate_pro_register_and_move() maps the PERSISTENT RESERVE OUT parameter list with transport_kmap_data_sg() and parses the destination TransportID with target_parse_pr_out_transport_id(). For an iSCSI TransportID (FORMAT CODE 01b), iscsi_parse_pr_out_transport_id() returns the ISID in iport_ptr as a raw pointer into that mapped buffer.

The function then unmaps the buffer with transport_kunmap_data_sg() before dereferencing iport_ptr in strcmp(), __core_scsi3_locate_pr_reg() and core_scsi3_alloc_registration(). When the parameter list spans more than one page (PARAMETER LIST LENGTH > 4096), transport_kmap_data_sg() uses vmap() and transport_kunmap_data_sg() does vunmap(), so the kernel virtual address backing iport_ptr is torn down and every subsequent dereference is a use-after-free read of the unmapped region.

Keep the parameter list mapped until iport_ptr is no longer needed: drop the early transport_kunmap_data_sg() and unmap once on the success path, right before returning. The error paths already unmap through the existing "if (buf) transport_kunmap_data_sg(cmd)" at the out: label, which now runs on every post-map error exit because buf is no longer cleared early. Only reads of the mapping happen while spinlocks are held; the map and unmap calls remain outside any lock. The sibling caller core_scsi3_decode_spec_i_port() already uses the buffer before unmapping it and is left unchanged.

Quoted 3 lines
Fixes 4949314c7283 ("target: Allow control CDBs with data > 1 page")

FWIW:

Reviewed-by on v3 1/1
#3 lore raw
In reply to Bryam Vargas
Quoted 99 lines · Bryam Vargas

core_scsi3_emulate_pro_register_and_move() maps the PERSISTENT RESERVE OUT parameter list with transport_kmap_data_sg() and parses the destination TransportID with target_parse_pr_out_transport_id(). For an iSCSI TransportID (FORMAT CODE 01b), iscsi_parse_pr_out_transport_id() returns the ISID in iport_ptr as a raw pointer into that mapped buffer.

The function then unmaps the buffer with transport_kunmap_data_sg() before dereferencing iport_ptr in strcmp(), __core_scsi3_locate_pr_reg() and core_scsi3_alloc_registration(). When the parameter list spans more than one page (PARAMETER LIST LENGTH > 4096), transport_kmap_data_sg() uses vmap() and transport_kunmap_data_sg() does vunmap(), so the kernel virtual address backing iport_ptr is torn down and every subsequent dereference is a use-after-free read of the unmapped region.

Keep the parameter list mapped until iport_ptr is no longer needed: drop the early transport_kunmap_data_sg() and unmap once on the success path, right before returning. The error paths already unmap through the existing "if (buf) transport_kunmap_data_sg(cmd)" at the out: label, which now runs on every post-map error exit because buf is no longer cleared early. Only reads of the mapping happen while spinlocks are held; the map and unmap calls remain outside any lock. The sibling caller core_scsi3_decode_spec_i_port() already uses the buffer before unmapping it and is left unchanged.

Fixes 4949314c7283 ("target: Allow control CDBs with data > 1 page")

---
v3 (review of v2 by John Garry and James Bottomley):
- Drop the parser-ownership approach. Rather than copy the ISID into an allocation that both callers must kfree() (v2), keep the PR-OUT parameter list mapped across the iport_ptr dereferences and move the single unmap to the success path. No allocation, no kfree, and target_core_fabric_lib.c / core_scsi3_decode_spec_i_port() are unchanged. This is the form John Garry asked for ("keep the mapping in place for longer, until the out: label") and removes the multiple kfree() paths v2 added.
- Re: holding the mapping while spinlocks are held (raised on v2): only reads of the mapping occur under dev_reservation_lock; the kmap/kunmap calls are all outside any lock, so there is no sleep-in-atomic concern.

v2: https://lore.kernel.org/linux-scsi/[email protected]/
v1: https://lore.kernel.org/linux-scsi/[email protected]/

Class / impact: CWE-416 use-after-free (use-after-vunmap) in the LIO SCSI target, reachable by an authenticated iSCSI initiator that is a current Persistent Reservation registrant on the LUN. It sends PERSISTENT RESERVE OUT / REGISTER AND MOVE with an iSCSI (FORMAT CODE 01b) TransportID and a PARAMETER LIST LENGTH > 4096 so the parameter list spans >1 page and is mapped with vmap(). After transport_kunmap_data_sg() vunmap()s that region, the retained iport_ptr is dereferenced -> kernel read of an unmapped vmalloc address (oops / DoS). Primarily a remotely reachable authenticated denial of service. Present in all maintained trees since 4949314c7283 (v3.3, 2012), which introduced the multi-page vmap() path. Verified at mainline v7.1-rc6 and stable v6.12.92.

Reproducer (authenticated iSCSI initiator, current PR reservation holder):
1. PERSISTENT RESERVE OUT / REGISTER a key from the iSCSI nexus.
2. PERSISTENT RESERVE OUT / REGISTER AND MOVE, FORMAT CODE 01b TransportID

     (IQN + ",i,0x" + 12-char ISID), RELATIVE TARGET PORT IDENTIFIER of an
     existing target port, with PARAMETER LIST LENGTH = 8192 (two pages ->
     vmap()/vunmap()), the inner ADDITIONAL LENGTH set so tid_len + 24 ==
     data_length, the remainder zero padding.

A/B verification (CONFIG_KASAN_VMALLOC=y, kasan.fault=report, x86-64, 6.12.90; same kernel for every arm; 64-bit and 32-bit initiator):
- Without this patch (8192-byte, two-page request):

      BUG: KASAN: vmalloc-out-of-bounds in strcmp+0xa7/0xb0
        strcmp
        core_scsi3_emulate_pro_register_and_move [target_core]
        ? remove_vm_area
        target_scsi3_emulate_pr_out [target_core]
        __target_execute_cmd / iscsit_execute_cmd / iscsi_target_rx_thread
      followed by "unable to handle page fault" (PTE 0); the command never
      completes (the iSCSI rx kthread dies).
  - Control (128-byte, single-page request): no report (kunmap is a no-op
    on 64-bit !HIGHMEM, no vunmap).
  - With this patch (same 8192-byte request, 64-bit and 32-bit initiator):
    no report, the command completes.
 drivers/target/target_core_pr.c | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)
drivers/target/target_core_pr.c +0 -3
index 11790f2c5d80..7c5bb7d67947 100644
@@ -3293,9 +3293,6 @@ core_scsi3_emulate_pro_register_and_move(struct se_cmd *cmd, u64 res_key,
 		goto out;
 	}
 
-	transport_kunmap_data_sg(cmd);
-	buf = NULL;
-
 	pr_debug("SPC-3 PR [%s] Extracted initiator %s identifier: %s"
 		" %s\n", dest_tf_ops->fabric_name, (iport_ptr != NULL) ?
 		"port" : "device", initiator_str, (iport_ptr != NULL) ?
@@ -3532,6 +3529,11 @@ core_scsi3_emulate_pro_register_and_move(struct se_cmd *cmd, u64 res_key,
 	core_scsi3_update_and_write_aptpl(cmd->se_dev, aptpl);

drivers/target/target_core_pr.c core_scsi3_emulate_pro_register_and_move(struct se_cmd *cmd, u64 res_key, · line 3539

Quoted 10 lines
drivers/target/target_core_pr.c +5 -0
 
 	core_scsi3_put_pr_reg(dest_pr_reg);
+	/*
+	 * iport_ptr aliases the PR-OUT parameter list mapped above, so the
+	 * buffer is unmapped only here on success (and at out: on error).
+	 */
+	transport_kunmap_data_sg(cmd);
 	return 0;
 out:
 	if (buf)

Looks good.