Message ID | 20200724213640.389191-13-keescook@chromium.org (mailing list archive) |
---|---|
State | Superseded |
Headers | show |
Series | Introduce partial kernel_read_file() support | expand |
On Fri, 2020-07-24 at 14:36 -0700, Kees Cook wrote: > Now that security_post_load_data() is wired up, use it instead > of the NULL file argument style of security_post_read_file(), > and update the security_kernel_load_data() call to indicate that a > security_kernel_post_load_data() call is expected. > > Wire up the IMA check to match earlier logic. Perhaps a generalized > change to ima_post_load_data() might look something like this: > > return process_buffer_measurement(buf, size, > kernel_load_data_id_str(load_id), > read_idmap[load_id] ?: FILE_CHECK, > 0, NULL); > > Signed-off-by: Kees Cook <keescook@chromium.org> process_measurement() measures, verifies a file signature - both signatures stored as an xattr and as an appended buffer signature - and augments audit records with the file hash. (Support for measuring, augmenting audit records, and/or verifying fs-verity signatures has yet to be added.) As explained in my response to 11/19, the file descriptor provides the file pathname associated with the buffer data. In addition, IMA policy rules may be defined in terms of other file descriptor info - uid, euid, uuid, etc. Recently support was added for measuring the kexec boot command line, certificates being loaded onto a keyring, and blacklisted file hashes (limited to appended signatures). None of these buffers are signed. process_buffer_measurement() was added for this reason and as a result is limited to just measuring the buffer data. Whether process_measurement() or process_buffer_measurement() should be modified, needs to be determined. In either case to support the init_module syscall, would at minimum require the associated file pathname. Mimi
On Mon, Jul 27, 2020 at 06:57:45AM -0400, Mimi Zohar wrote: > On Fri, 2020-07-24 at 14:36 -0700, Kees Cook wrote: > > Now that security_post_load_data() is wired up, use it instead > > of the NULL file argument style of security_post_read_file(), > > and update the security_kernel_load_data() call to indicate that a > > security_kernel_post_load_data() call is expected. > > > > Wire up the IMA check to match earlier logic. Perhaps a generalized > > change to ima_post_load_data() might look something like this: > > > > return process_buffer_measurement(buf, size, > > kernel_load_data_id_str(load_id), > > read_idmap[load_id] ?: FILE_CHECK, > > 0, NULL); > > > > Signed-off-by: Kees Cook <keescook@chromium.org> > > process_measurement() measures, verifies a file signature - both > signatures stored as an xattr and as an appended buffer signature - > and augments audit records with the file hash. (Support for measuring, > augmenting audit records, and/or verifying fs-verity signatures has > yet to be added.) > > As explained in my response to 11/19, the file descriptor provides the > file pathname associated with the buffer data. In addition, IMA > policy rules may be defined in terms of other file descriptor info - > uid, euid, uuid, etc. > > Recently support was added for measuring the kexec boot command line, > certificates being loaded onto a keyring, and blacklisted file hashes > (limited to appended signatures). None of these buffers are signed. > process_buffer_measurement() was added for this reason and as a > result is limited to just measuring the buffer data. > > Whether process_measurement() or process_buffer_measurement() should > be modified, needs to be determined. In either case to support the > init_module syscall, would at minimum require the associated file > pathname. Right -- I don't intend to make changes to the init_module() syscall since it's deprecated, so this hook is more of a "fuller LSM coverage for old syscalls" addition. IMA can happily continue to ignore it, which is what I have here, but I thought I'd at least show what it *might* look like. Perhaps BPF LSM is a better example. Does anything need to change for this patch?
On Tue, 2020-07-28 at 12:43 -0700, Kees Cook wrote: > On Mon, Jul 27, 2020 at 06:57:45AM -0400, Mimi Zohar wrote: > > On Fri, 2020-07-24 at 14:36 -0700, Kees Cook wrote: > > > Now that security_post_load_data() is wired up, use it instead > > > of the NULL file argument style of security_post_read_file(), > > > and update the security_kernel_load_data() call to indicate that a > > > security_kernel_post_load_data() call is expected. > > > > > > Wire up the IMA check to match earlier logic. Perhaps a generalized > > > change to ima_post_load_data() might look something like this: > > > > > > return process_buffer_measurement(buf, size, > > > kernel_load_data_id_str(load_id), > > > read_idmap[load_id] ?: FILE_CHECK, > > > 0, NULL); > > > > > > Signed-off-by: Kees Cook <keescook@chromium.org> > > > > process_measurement() measures, verifies a file signature - both > > signatures stored as an xattr and as an appended buffer signature - > > and augments audit records with the file hash. (Support for measuring, > > augmenting audit records, and/or verifying fs-verity signatures has > > yet to be added.) > > > > As explained in my response to 11/19, the file descriptor provides the > > file pathname associated with the buffer data. In addition, IMA > > policy rules may be defined in terms of other file descriptor info - > > uid, euid, uuid, etc. > > > > Recently support was added for measuring the kexec boot command line, > > certificates being loaded onto a keyring, and blacklisted file hashes > > (limited to appended signatures). None of these buffers are signed. > > process_buffer_measurement() was added for this reason and as a > > result is limited to just measuring the buffer data. > > > > Whether process_measurement() or process_buffer_measurement() should > > be modified, needs to be determined. In either case to support the > > init_module syscall, would at minimum require the associated file > > pathname. > > Right -- I don't intend to make changes to the init_module() syscall > since it's deprecated, so this hook is more of a "fuller LSM coverage > for old syscalls" addition. > > IMA can happily continue to ignore it, which is what I have here, but I > thought I'd at least show what it *might* look like. Perhaps BPF LSM is > a better example. > > Does anything need to change for this patch? I wasn't aware that init_syscall was deprecated. From your original comments, it sounded like you wanted a new LSM for verifying kernel module signatures, as they're currently supported via init_module(). I was mistaken. Without a file descriptor, security_post_load_data() will measure the firmware, as Scott confirmed, but won't be able to verify the signature, whether he signed it using evmctl or not. Mimi
On Wed, 2020-07-29 at 12:29 -0400, Mimi Zohar wrote: > On Tue, 2020-07-28 at 12:43 -0700, Kees Cook wrote: > > On Mon, Jul 27, 2020 at 06:57:45AM -0400, Mimi Zohar wrote: > > > On Fri, 2020-07-24 at 14:36 -0700, Kees Cook wrote: > > > > Now that security_post_load_data() is wired up, use it instead > > > > of the NULL file argument style of security_post_read_file(), > > > > and update the security_kernel_load_data() call to indicate that a > > > > security_kernel_post_load_data() call is expected. > > > > > > > > Wire up the IMA check to match earlier logic. Perhaps a generalized > > > > change to ima_post_load_data() might look something like this: > > > > > > > > return process_buffer_measurement(buf, size, > > > > kernel_load_data_id_str(load_id), > > > > read_idmap[load_id] ?: FILE_CHECK, > > > > 0, NULL); > > > > > > > > Signed-off-by: Kees Cook <keescook@chromium.org> > > > > > > process_measurement() measures, verifies a file signature - both > > > signatures stored as an xattr and as an appended buffer signature - > > > and augments audit records with the file hash. (Support for measuring, > > > augmenting audit records, and/or verifying fs-verity signatures has > > > yet to be added.) > > > > > > As explained in my response to 11/19, the file descriptor provides the > > > file pathname associated with the buffer data. In addition, IMA > > > policy rules may be defined in terms of other file descriptor info - > > > uid, euid, uuid, etc. > > > > > > Recently support was added for measuring the kexec boot command line, > > > certificates being loaded onto a keyring, and blacklisted file hashes > > > (limited to appended signatures). None of these buffers are signed. > > > process_buffer_measurement() was added for this reason and as a > > > result is limited to just measuring the buffer data. > > > > > > Whether process_measurement() or process_buffer_measurement() should > > > be modified, needs to be determined. In either case to support the > > > init_module syscall, would at minimum require the associated file > > > pathname. > > > > Right -- I don't intend to make changes to the init_module() syscall > > since it's deprecated, so this hook is more of a "fuller LSM coverage > > for old syscalls" addition. > > > > IMA can happily continue to ignore it, which is what I have here, but I > > thought I'd at least show what it *might* look like. Perhaps BPF LSM is > > a better example. > > > > Does anything need to change for this patch? > > I wasn't aware that init_syscall was deprecated. From your original comments, > it sounded like you wanted a new LSM for verifying kernel module signatures, > as > they're currently supported via init_module(). > > I was mistaken. Without a file descriptor, security_post_load_data() will > measure the firmware, as Scott confirmed, but won't be able to verify the > signature, whether he signed it using evmctl or not, Actually, the partial firmware read should be calling security_kernel_read_file(). The sysfs firmware fallback is calling security_kernel_load_data(). Which firmware is calling security_kernel_post_load_data()? thanks, Mimi
On Wed, Jul 29, 2020 at 02:10:18PM -0400, Mimi Zohar wrote: > Actually, the partial firmware read should be calling > security_kernel_read_file(). Yup, it does[1], and when "whole_file" is true, it will call security_kernel_post_read_file() with the buffer contents at the end. > The sysfs firmware fallback is calling security_kernel_load_data(). Correct[2]; it has no file associated with it (same as the EFI platform source). > Which firmware is calling security_kernel_post_load_data()? sysfs and platform both call it[2], matched with their security_kernel_load_data() calls. -Kees [1] v4 patch 14: "fs/kernel_file_read: Add "offset" arg for partial reads" https://lore.kernel.org/lkml/20200729175845.1745471-1-keescook@chromium.org/T/#iZ2e.:..:20200729175845.1745471-15-keescook::40chromium.org:0fs:kernel_read_file.c [2] v4 patch 10: "firmware_loader: Use security_post_load_data()" https://lore.kernel.org/lkml/20200729175845.1745471-1-keescook@chromium.org/T/#iZ2e.:..:20200729175845.1745471-11-keescook::40chromium.org:0drivers:base:firmware_loader:fallback.c
diff --git a/drivers/base/firmware_loader/fallback.c b/drivers/base/firmware_loader/fallback.c index a196aacce22c..7cfdfdcb819c 100644 --- a/drivers/base/firmware_loader/fallback.c +++ b/drivers/base/firmware_loader/fallback.c @@ -272,9 +272,9 @@ static ssize_t firmware_loading_store(struct device *dev, dev_err(dev, "%s: map pages failed\n", __func__); else - rc = security_kernel_post_read_file(NULL, - fw_priv->data, fw_priv->size, - READING_FIRMWARE); + rc = security_kernel_post_load_data(fw_priv->data, + fw_priv->size, + LOADING_FIRMWARE); /* * Same logic as fw_load_abort, only the DONE bit @@ -613,7 +613,7 @@ static bool fw_run_sysfs_fallback(u32 opt_flags) return false; /* Also permit LSMs and IMA to fail firmware sysfs fallback */ - ret = security_kernel_load_data(LOADING_FIRMWARE, false); + ret = security_kernel_load_data(LOADING_FIRMWARE, true); if (ret < 0) return false; diff --git a/drivers/base/firmware_loader/fallback_platform.c b/drivers/base/firmware_loader/fallback_platform.c index a12c79d47efc..4d1157af0e86 100644 --- a/drivers/base/firmware_loader/fallback_platform.c +++ b/drivers/base/firmware_loader/fallback_platform.c @@ -17,7 +17,7 @@ int firmware_fallback_platform(struct fw_priv *fw_priv, u32 opt_flags) if (!(opt_flags & FW_OPT_FALLBACK_PLATFORM)) return -ENOENT; - rc = security_kernel_load_data(LOADING_FIRMWARE, false); + rc = security_kernel_load_data(LOADING_FIRMWARE, true); if (rc) return rc; @@ -27,6 +27,11 @@ int firmware_fallback_platform(struct fw_priv *fw_priv, u32 opt_flags) if (fw_priv->data && size > fw_priv->allocated_size) return -ENOMEM; + + rc = security_kernel_post_load_data((u8 *)data, size, LOADING_FIRMWARE); + if (rc) + return rc; + if (!fw_priv->data) fw_priv->data = vmalloc(size); if (!fw_priv->data) diff --git a/security/integrity/ima/ima_main.c b/security/integrity/ima/ima_main.c index 85000dc8595c..1a7bc4c7437d 100644 --- a/security/integrity/ima/ima_main.c +++ b/security/integrity/ima/ima_main.c @@ -648,15 +648,6 @@ int ima_post_read_file(struct file *file, void *buf, loff_t size, enum ima_hooks func; u32 secid; - if (!file && read_id == READING_FIRMWARE) { - if ((ima_appraise & IMA_APPRAISE_FIRMWARE) && - (ima_appraise & IMA_APPRAISE_ENFORCE)) { - pr_err("Prevent firmware loading_store.\n"); - return -EACCES; /* INTEGRITY_UNKNOWN */ - } - return 0; - } - /* permit signed certs */ if (!file && read_id == READING_X509_CERTIFICATE) return 0; @@ -706,7 +697,7 @@ int ima_load_data(enum kernel_load_data_id id, bool contents) } break; case LOADING_FIRMWARE: - if (ima_enforce && (ima_appraise & IMA_APPRAISE_FIRMWARE)) { + if (ima_enforce && (ima_appraise & IMA_APPRAISE_FIRMWARE) && !contents) { pr_err("Prevent firmware sysfs fallback loading.\n"); return -EACCES; /* INTEGRITY_UNKNOWN */ } @@ -739,6 +730,15 @@ int ima_load_data(enum kernel_load_data_id id, bool contents) */ int ima_post_load_data(char *buf, loff_t size, enum kernel_load_data_id load_id) { + if (load_id == LOADING_FIRMWARE) { + if ((ima_appraise & IMA_APPRAISE_FIRMWARE) && + (ima_appraise & IMA_APPRAISE_ENFORCE)) { + pr_err("Prevent firmware loading_store.\n"); + return -EACCES; /* INTEGRITY_UNKNOWN */ + } + return 0; + } + return 0; }
Now that security_post_load_data() is wired up, use it instead of the NULL file argument style of security_post_read_file(), and update the security_kernel_load_data() call to indicate that a security_kernel_post_load_data() call is expected. Wire up the IMA check to match earlier logic. Perhaps a generalized change to ima_post_load_data() might look something like this: return process_buffer_measurement(buf, size, kernel_load_data_id_str(load_id), read_idmap[load_id] ?: FILE_CHECK, 0, NULL); Signed-off-by: Kees Cook <keescook@chromium.org> --- drivers/base/firmware_loader/fallback.c | 8 ++++---- .../base/firmware_loader/fallback_platform.c | 7 ++++++- security/integrity/ima/ima_main.c | 20 +++++++++---------- 3 files changed, 20 insertions(+), 15 deletions(-)