Xen Live Patching Design v1

Rationale

A mechanism is required to binarily patch the running hypervisor with new opcodes that have come about due to primarily security updates.

This document describes the design of the API that would allow us to upload to the hypervisor binary patches.

The document is split in four sections:

Glossary

History

The document has gone under various reviews and only covers v1 design.

The end of the document has a section titled Not Yet Done which outlines ideas and design for the future version of this work.

Multiple ways to patch

The mechanism needs to be flexible to patch the hypervisor in multiple ways and be as simple as possible. The compiled code is contiguous in memory with no gaps - so we have no luxury of 'moving' existing code and must either insert a trampoline to the new code to be executed - or only modify in-place the code if there is sufficient space. The placement of new code has to be done by hypervisor and the virtual address for the new code is allocated dynamically.

This implies that the hypervisor must compute the new offsets when splicing in the new trampoline code. Where the trampoline is added (inside the function we are patching or just the callers?) is also important.

To lessen the amount of code in hypervisor, the consumer of the API is responsible for identifying which mechanism to employ and how many locations to patch. Combinations of modifying in-place code, adding trampoline, etc has to be supported. The API should allow read/write any memory within the hypervisor virtual address space.

We must also have a mechanism to query what has been applied and a mechanism to revert it if needed.

Workflow

The expected workflows of higher-level tools that manage multiple patches on production machines would be:

Patching code

The first mechanism to patch that comes in mind is in-place replacement. That is replace the affected code with new code. Unfortunately the x86 ISA is variable size which places limits on how much space we have available to replace the instructions. That is not a problem if the change is smaller than the original opcode and we can fill it with nops. Problems will appear if the replacement code is longer.

The second mechanism is by ti replace the call or jump to the old function with the address of the new function.

A third mechanism is to add a jump to the new function at the start of the old function. N.B. The Xen hypervisor implements the third mechanism. See Trampoline (e9 opcode) section for more details.

Example of trampoline and in-place splicing

As example we will assume the hypervisor does not have XSA-132 (see domctl/sysctl: don't leak hypervisor stack to toolstacks 4ff3449f0e9d175ceb9551d3f2aecb59273f639d) and we would like to binary patch the hypervisor with it. The original code looks as so:

   48 89 e0                  mov    %rsp,%rax  
   48 25 00 80 ff ff         and    $0xffffffffffff8000,%rax  

while the new patched hypervisor would be:

   48 c7 45 b8 00 00 00 00   movq   $0x0,-0x48(%rbp)  
   48 c7 45 c0 00 00 00 00   movq   $0x0,-0x40(%rbp)  
   48 c7 45 c8 00 00 00 00   movq   $0x0,-0x38(%rbp)  
   48 89 e0                  mov    %rsp,%rax  
   48 25 00 80 ff ff         and    $0xffffffffffff8000,%rax  

This is inside the archdodomctl. This new change adds 21 extra bytes of code which alters all the offsets inside the function. To alter these offsets and add the extra 21 bytes of code we might not have enough space in .text to squeeze this in.

As such we could simplify this problem by only patching the site which calls archdodomctl:

do_domctl:  
 e8 4b b1 05 00          callq  ffff82d08015fbb9   

with a new address for where the new arch_do_domctl would be (this area would be allocated dynamically).

Astute readers will wonder what we need to do if we were to patch do_domctl - which is not called directly by hypervisor but on behalf of the guests via the compat_hypercall_table and hypercall_table. Patching the offset in hypercall_table for `do_domctl: (ffff82d080103079 :)


 ffff82d08024d490:   79 30  
 ffff82d08024d492:   10 80 d0 82 ff ff   

with the new address where the new do_domctl is possible. The other place where it is used is in hvm_hypercall64_table which would need to be patched in a similar way. This would require an in-place splicing of the new virtual address of arch_do_domctl.

In summary this example patched the callee of the affected function by * allocating memory for the new code to live in, * changing the virtual address in all the functions which called the old code (computing the new offset, patching the callq with a new callq). * changing the function pointer tables with the new virtual address of the function (splicing in the new virtual address). Since this table resides in the .rodata section we would need to temporarily change the page table permissions during this part.

However it has drawbacks - the safety checks which have to make sure the function is not on the stack - must also check every caller. For some patches this could mean - if there were an sufficient large amount of callers - that we would never be able to apply the update.

Having the patching done at predetermined instances where the stacks are not deep mostly solves this problem.

Example of different trampoline patching.

An alternative mechanism exists where we can insert a trampoline in the existing function to be patched to jump directly to the new code. This lessens the locations to be patched to one but it puts pressure on the CPU branching logic (I-cache, but it is just one unconditional jump).

For this example we will assume that the hypervisor has not been compiled with fe2e079f642effb3d24a6e1a7096ef26e691d93e (XSA-125: pre-fill structures for certain HYPERVISOR_xen_version sub-ops) which mem-sets an structure in xen_version hypercall. This function is not called anywhere in the hypervisor (it is called by the guest) but referenced in the compat_hypercall_table and hypercall_table (and indirectly called from that). Patching the offset in hypercall_table for the old do_xen_version (ffff82d080112f9e )

ffff82d08024b270 :
...
ffff82d08024b2f8: 9e 2f 11 80 d0 82 ff ff

with the new address where the new do_xen_version is possible. The other place where it is used is in hvm_hypercall64_table which would need to be patched in a similar way. This would require an in-place splicing of the new virtual address of do_xen_version.

An alternative solution would be to patch insert a trampoline in the old do_xen_version' function to directly jump to the newdoxenversion`.

 ffff82d080112f9e do_xen_version:  
 ffff82d080112f9e:       48 c7 c0 da ff ff ff    mov    $0xffffffffffffffda,%rax  
 ffff82d080112fa5:       83 ff 09                cmp    $0x9,%edi  
 ffff82d080112fa8:       0f 87 24 05 00 00       ja     ffff82d0801134d2 ; do_xen_version+0x534  

with:

 ffff82d080112f9e do_xen_version:  
 ffff82d080112f9e:       e9 XX YY ZZ QQ          jmpq   [new do_xen_version]  

which would lessen the amount of patching to just one location.

In summary this example patched the affected function to jump to the new replacement function which required: * allocating memory for the new code to live in, * inserting trampoline with new offset in the old function to point to the new function. * Optionally we can insert in the old function a trampoline jump to an function providing an BUG_ON to catch errant code.

The disadvantage of this are that the unconditional jump will consume a small I-cache penalty. However the simplicity of the patching and higher chance of passing safety checks make this a worthwhile option.

This patching has a similar drawback as inline patching - the safety checks have to make sure the function is not on the stack. However since we are replacing at a higher level (a full function as opposed to various offsets within functions) the checks are simpler.

Having the patching done at predetermined instances where the stacks are not deep mostly solves this problem as well.

Security

With this method we can re-write the hypervisor - and as such we MUST be diligent in only allowing certain guests to perform this operation.

Furthermore with SecureBoot or tboot, we MUST also verify the signature of the payload to be certain it came from a trusted source and integrity was intact.

As such the hypercall MUST support an XSM policy to limit what the guest is allowed to invoke. If the system is booted with signature checking the signature checking will be enforced.

Design of payload format

The payload MUST contain enough data to allow us to apply the update and also safely reverse it. As such we MUST know:

This binary format can be constructed using an custom binary format but there are severe disadvantages of it:

As such having the payload in an ELF file is the sensible way. We would be carrying the various sets of structures (and data) in the ELF sections under different names and with definitions.

Note that every structure has padding. This is added so that the hypervisor can re-use those fields as it sees fit.

Earlier design attempted to ineptly explain the relations of the ELF sections to each other without using proper ELF mechanism (shinfo, shlink, data structures using Elf types, etc). This design will explain the structures and how they are used together and not dig in the ELF format - except mention that the section names should match the structure names.

The Xen Live Patch payload is a relocatable ELF binary. A typical binary would have:

It may also have some architecture-specific sections. For example:

The Xen Live Patch core code loads the payload as a standard ELF binary, relocates it and handles the architecture-specifc sections as needed. This process is much like what the Linux kernel module loader does.

The payload contains at least three sections:

.livepatch.funcs

The .livepatch.funcs contains an array of livepatch_func structures which describe the functions to be patched:

struct livepatch_func {  
    const char *name;  
    void *new_addr;  
    void *old_addr;  
    uint32_t new_size;  
    uint32_t old_size;  
    uint8_t version;  
    uint8_t opaque[31];  
};  

The size of the structure is 64 bytes on 64-bit hypervisors. It will be 52 on 32-bit hypervisors.

The size of the livepatch_func array is determined from the ELF section size.

When applying the patch the hypervisor iterates over each livepatch_func structure and the core code inserts a trampoline at old_addr to new_addr. The new_addr is altered when the ELF payload is loaded.

When reverting a patch, the hypervisor iterates over each livepatch_func and the core code copies the data from the undo buffer (private internal copy) to old_addr.

It optionally may contain the address of functions to be called right before being applied and after being reverted:

Example of .livepatch.funcs

A simple example of what a payload file can be:

/* MUST be in sync with hypervisor. */  
struct livepatch_func {  
    const char *name;  
    void *new_addr;  
    void *old_addr;  
    uint32_t new_size;  
    uint32_t old_size;  
    uint8_t version;
    uint8_t pad[31];  
};  

/* Our replacement function for xen_extra_version. */  
const char *xen_hello_world(void)  
{  
    return "Hello World";  
}  

static unsigned char patch_this_fnc[] = "xen_extra_version";  

struct livepatch_func livepatch_hello_world = {  
    .version = LIVEPATCH_PAYLOAD_VERSION,
    .name = patch_this_fnc,  
    .new_addr = xen_hello_world,  
    .old_addr = (void *)0xffff82d08013963c, /* Extracted from xen-syms. */  
    .new_size = 13, /* To be be computed by scripts. */  
    .old_size = 13, /* -----------""---------------  */  
} __attribute__((__section__(".livepatch.funcs")));  

Code must be compiled with -fPIC.

.livepatch.hooks.load and .livepatch.hooks.unload

This section contains an array of function pointers to be executed before payload is being applied (.livepatch.funcs) or after reverting the payload. This is useful to prepare data structures that need to be modified patching.

Each entry in this array is eight bytes.

The type definition of the function are as follow:

typedef void (*livepatch_loadcall_t)(void);  
typedef void (*livepatch_unloadcall_t)(void);   

.livepatch.depends and .note.gnu.build-id

To support dependencies checking and safe loading (to load the appropiate payload against the right hypervisor) there is a need to embbed an build-id dependency.

This is done by the payload containing an section .livepatch.depends which follows the format of an ELF Note. The contents of this (name, and description) are specific to the linker utilized to build the hypevisor and payload.

If GNU linker is used then the name is GNU and the description is a NTGNUBUILD_ID type ID. The description can be an SHA1 checksum, MD5 checksum or any unique value.

The size of these structures varies with the --build-id linker option.

Hypercalls

We will employ the sub operations of the system management hypercall (sysctl). There are to be four sub-operations:

Most of the actions are asynchronous therefore the caller is responsible to verify that it has been applied properly by retrieving the summary of it and verifying that there are no error codes associated with the payload.

We MUST make some of them asynchronous due to the nature of patching it requires every physical CPU to be lock-step with each other. The patching mechanism while an implementation detail, is not an short operation and as such the design MUST assume it will be an long-running operation.

The sub-operations will spell out how preemption is to be handled (if at all).

Furthermore it is possible to have multiple different payloads for the same function. As such an unique name per payload has to be visible to allow proper manipulation.

The hypercall is part of the xen_sysctl. The top level structure contains one uint32_t to determine the sub-operations and one padding field which MUST always be zero.

struct xen_sysctl_livepatch_op {  
    uint32_t cmd;                   /* IN: XEN_SYSCTL_LIVEPATCH_*. */  
    uint32_t pad;                   /* IN: Always zero. */  
    union {  
          ... see below ...  
        } u;  
};  

while the rest of hypercall specific structures are part of the this structure.

Basic type: struct xenlivepatchname

Most of the hypercalls employ an shared structure called struct xen_livepatch_name which contains:

The structure is as follow:

/*  
 *  Uniquely identifies the payload.  Should be human readable.  
 * Includes the NUL terminator  
 */  
#define XEN_LIVEPATCH_NAME_SIZE 128  
struct xen_livepatch_name {  
    XEN_GUEST_HANDLE_64(char) name;         /* IN, pointer to name. */  
    uint16_t size;                          /* IN, size of name. May be upto   
                                               XEN_LIVEPATCH_NAME_SIZE. */  
    uint16_t pad[3];                        /* IN: MUST be zero. */ 
};  

XENSYSCTLLIVEPATCH_UPLOAD (0)

Upload a payload to the hypervisor. The payload is verified against basic checks and if there are any issues the proper return code will be returned. The payload is not applied at this time - that is controlled by XEN_SYSCTL_LIVEPATCH_ACTION.

The caller provides:

The name could be an UUID that stays fixed forever for a given payload. It can be embedded into the ELF payload at creation time and extracted by tools.

The return value is zero if the payload was succesfully uploaded. Otherwise an -XEN_EXX return value is provided. Duplicate name are not supported.

The payload is the ELF payload as mentioned in the Payload format section.

The structure is as follow:

struct xen_sysctl_livepatch_upload {  
    xen_livepatch_name_t name;          /* IN, name of the patch. */  
    uint64_t size;                      /* IN, size of the ELF file. */  
    XEN_GUEST_HANDLE_64(uint8) payload; /* IN: ELF file. */  
};  

XENSYSCTLLIVEPATCH_GET (1)

Retrieve an status of an specific payload. This caller provides:

Upon completion the struct xen_livepatch_status is updated.

The return value of the hypercall is zero on success and -XENEXX on failure. (Note that the `rc`` value can be different from the return value, as in rc=-XENEAGAIN and return value can be 0).

For example, supposing there is an payload:

 status: LIVEPATCH_STATUS_CHECKED
 rc: 0

We apply an action - LIVEPATCHACTIONREVERT - to revert it (which won't work as we have not even applied it. Afterwards we will have:

 status: LIVEPATCH_STATUS_CHECKED
 rc: -XEN_EINVAL

It has failed but it remains loaded.

This operation is synchronous and does not require preemption.

The structure is as follow:

struct xen_livepatch_status {  
#define LIVEPATCH_STATUS_CHECKED      1  
#define LIVEPATCH_STATUS_APPLIED      2  
    uint32_t state;                 /* OUT: LIVEPATCH_STATE_*. */  
    int32_t rc;                     /* OUT: 0 if no error, otherwise -XEN_EXX. */  
};  

struct xen_sysctl_livepatch_get {  
    xen_livepatch_name_t name;      /* IN, the name of the payload. */  
    xen_livepatch_status_t status;  /* IN/OUT: status of the payload. */  
};  

XENSYSCTLLIVEPATCH_LIST (2)

Retrieve an array of abbreviated status and names of payloads that are loaded in the hypervisor.

The caller provides:

If the hypercall returns an positive number, it is the number (upto nr provided to the hypercall) of the payloads returned, along with nr updated with the number of remaining payloads, version updated (it may be the same across hypercalls - if it varies the data is stale and further calls could fail). The status, name, and len' are updated at their designed index value (idx) with the returned value of data.

If the hypercall returns -XEN_E2BIG the nr is too big and should be lowered.

If the hypercall returns an zero value there are no more payloads.

Note that due to the asynchronous nature of hypercalls the control domain might have added or removed a number of payloads making this information stale. It is the responsibility of the toolstack to use the version field to check between each invocation. if the version differs it should discard the stale data and start from scratch. It is OK for the toolstack to use the new version field.

The struct xen_livepatch_status structure contains an status of payload which includes:

The structure is as follow:

struct xen_sysctl_livepatch_list {  
    uint32_t version;                       /* OUT: Hypervisor stamps value.
                                               If varies between calls, we are  
                                               getting stale data. */  
    uint32_t idx;                           /* IN: Index into hypervisor list. */
    uint32_t nr;                            /* IN: How many status, names, and len  
                                               should be filled out. Can be zero to get  
                                               amount of payloads and version.  
                                               OUT: How many payloads left. */  
    uint32_t pad;                           /* IN: Must be zero. */  
    XEN_GUEST_HANDLE_64(xen_livepatch_status_t) status;  /* OUT. Must have enough  
                                               space allocate for nr of them. */  
    XEN_GUEST_HANDLE_64(char) id;           /* OUT: Array of names. Each member  
                                               MUST XEN_LIVEPATCH_NAME_SIZE in size.  
                                               Must have nr of them. */  
    XEN_GUEST_HANDLE_64(uint32) len;        /* OUT: Array of lengths of name's.  
                                               Must have nr of them. */  
};  

XENSYSCTLLIVEPATCH_ACTION (3)

Perform an operation on the payload structure referenced by the name field. The operation request is asynchronous and the status should be retrieved by using either XENSYSCTLLIVEPATCHGET or XENSYSCTLLIVEPATCHLIST hypercall.

The caller provides:

The return value will be zero unless the provided fields are incorrect.

The structure is as follow:

#define LIVEPATCH_ACTION_UNLOAD  1  
#define LIVEPATCH_ACTION_REVERT  2  
#define LIVEPATCH_ACTION_APPLY   3  
#define LIVEPATCH_ACTION_REPLACE 4  
struct xen_sysctl_livepatch_action {  
    xen_livepatch_name_t name;              /* IN, name of the patch. */  
    uint32_t cmd;                           /* IN: LIVEPATCH_ACTION_* */  
    uint32_t time;                          /* IN: Zero if no timeout. */   
                                            /* Or upper bound of time (ms) */   
                                            /* for operation to take. */  
};  

State diagrams of LIVEPATCH_ACTION commands.

There is a strict ordering state of what the commands can be. The LIVEPATCHACTION prefix has been dropped to easy reading and does not include the LIVEPATCHSTATES:

              /->\  
              \  /  
 UNLOAD <--- CHECK ---> REPLACE|APPLY --> REVERT --\  
                \                                  |  
                 \-------------------<-------------/  

State transition table of LIVEPATCHACTION commands and LIVEPATCHSTATUS.

Note that:

The state transition table of valid states and action states:


+---------+---------+--------------------------------+-------+--------+
| ACTION  | Current | Result                         | Next STATE:    |
| ACTION  | STATE   |                                |CHECKED|APPLIED |
+---------+----------+-------------------------------+-------+--------+
| UNLOAD  | CHECKED | Unload payload. Always works.  |       |        |
|         |         | No next states.                |       |        |
+---------+---------+--------------------------------+-------+--------+
| APPLY   | CHECKED | Apply payload (success).       |       |   x    |
+---------+---------+--------------------------------+-------+--------+
| APPLY   | CHECKED | Apply payload (error|timeout)  |   x   |        |
+---------+---------+--------------------------------+-------+--------+
| REPLACE | CHECKED | Revert payloads and apply new  |       |   x    |
|         |         | payload with success.          |       |        |
+---------+---------+--------------------------------+-------+--------+
| REPLACE | CHECKED | Revert payloads and apply new  |   x   |        |
|         |         | payload with error.            |       |        |
+---------+---------+--------------------------------+-------+--------+
| REVERT  | APPLIED | Revert payload (success).      |   x   |        |
+---------+---------+--------------------------------+-------+--------+
| REVERT  | APPLIED | Revert payload (error|timeout) |       |   x    |
+---------+---------+--------------------------------+-------+--------+

All the other state transitions are invalid.

Sequence of events.

The normal sequence of events is to:

  1. XEN_SYSCTL_LIVEPATCH_UPLOAD to upload the payload. If there are errors STOP here.
  2. XEN_SYSCTL_LIVEPATCH_GET to check the ->rc. If -XEN_EAGAIN spin. If zero go to next step.
  3. XEN_SYSCTL_LIVEPATCH_ACTION with LIVEPATCH_ACTION_APPLY to apply the patch.
  4. XEN_SYSCTL_LIVEPATCH_GET to check the ->rc. If in -XEN_EAGAIN spin. If zero exit with success.

Addendum

Implementation quirks should not be discussed in a design document.

However these observations can provide aid when developing against this document.

Alternative assembler

Alternative assembler is a mechanism to use different instructions depending on what the CPU supports. This is done by providing multiple streams of code that can be patched in - or if the CPU does not support it - padded with nop operations. The alternative assembler macros cause the compiler to expand the code to place a most generic code in place - emit a special ELF .section header to tag this location. During run-time the hypervisor can leave the areas alone or patch them with an better suited opcodes.

Note that patching functions that copy to or from guest memory requires to support alternative support. For example this can be due to SMAP (specifically stac and clac operations) which is enabled on Broadwell and later architectures. It may be related to other alternative instructions.

When to patch

During the discussion on the design two candidates bubbled where the call stack for each CPU would be deterministic. This would minimize the chance of the patch not being applied due to safety checks failing. Safety checks such as not patching code which is on the stack - which can lead to corruption.

Rendezvous code instead of stop_machine for patching

The hypervisor's time rendezvous code runs synchronously across all CPUs every second. Using the stop_machine to patch can stall the time rendezvous code and result in NMI. As such having the patching be done at the tail of rendezvous code should avoid this problem.

However the entrance point for that code is dosoftirq->timersoftirqaction->timecalibration which ends up calling onselectedcpus on remote CPUs.

The remote CPUs receive CALLFUNCTIONVECTOR IPI and execute the desired function.

Before entering the guest code.

Before we call VMXResume we check whether any soft IRQs need to be executed. This is a good spot because all Xen stacks are effectively empty at that point.

To randezvous all the CPUs an barrier with an maximum timeout (which could be adjusted), combined with forcing all other CPUs through the hypervisor with IPIs, can be utilized to execute lockstep instructions on all CPUs.

The approach is similar in concept to stop_machine and the time rendezvous but is time-bound. However the local CPU stack is much shorter and a lot more deterministic.

This is implemented in the Xen Project hypervisor.

Compiling the hypervisor code

Hotpatch generation often requires support for compiling the target with -ffunction-sections / -fdata-sections. Changes would have to be done to the linker scripts to support this.

Generation of Live Patch ELF payloads

The design of that is not discussed in this design.

This is implemented in a seperate tool which lives in a seperate GIT repo.

Currently it resides at git://xenbits.xen.org/livepatch-build-tools.git

Exception tables and symbol tables growth

We may need support for adapting or augmenting exception tables if patching such code. Hotpatches may need to bring their own small exception tables (similar to how Linux modules support this).

If supporting hotpatches that introduce additional exception-locations is not important, one could also change the exception table in-place and reorder it afterwards.

As found almost every patch (XSA) to a non-trivial function requires additional entries in the exception table and/or the bug frames.

This is implemented in the Xen Project hypervisor.

.rodata sections

The patching might require strings to be updated as well. As such we must be also able to patch the strings as needed. This sounds simple - but the compiler has a habit of coalescing strings that are the same - which means if we in-place alter the strings - other users will be inadvertently affected as well.

This is also where pointers to functions live - and we may need to patch this as well. And switch-style jump tables.

To guard against that we must be prepared to do patching similar to trampoline patching or in-line depending on the flavour. If we can do in-line patching we would need to:

If are doing trampoline patching we would need to:

The trampoline patching is implemented in the Xen Project hypervisor.

.bss and .data sections.

In place patching writable data is not suitable as it is unclear what should be done depending on the current state of data. As such it should not be attempted.

However, functions which are being patched can bring in changes to strings (.data or .rodata section changes), or even to .bss sections.

As such the ELF payload can introduce new .rodata, .bss, and .data sections. Patching in the new function will end up also patching in the new .rodata section and the new function will reference the new string in the new .rodata section.

This is implemented in the Xen Project hypervisor.

Security

Only the privileged domain should be allowed to do this operation.

Live patch interdependencies

Live patch patches interdependencies are tricky.

There are the ways this can be addressed: * A single large patch that subsumes and replaces all previous ones. Over the life-time of patching the hypervisor this large patch grows to accumulate all the code changes. * Hotpatch stack - where an mechanism exists that loads the hotpatches in the same order they were built in. We would need an build-id of the hypevisor to make sure the hot-patches are build against the correct build. * Payload containing the old code to check against that. That allows the hotpatches to be loaded indepedently (if they don't overlap) - or if the old code also containst previously patched code - even if they overlap.

The disadvantage of the first large patch is that it can grow over time and not provide an bisection mechanism to identify faulty patches.

The hot-patch stack puts stricts requirements on the order of the patches being loaded and requires an hypervisor build-id to match against.

The old code allows much more flexibility and an additional guard, but is more complex to implement.

The second option which requires an build-id of the hypervisor is implemented in the Xen Project hypervisor.

Specifically each payload has two build-id ELF notes: * The build-id of the payload itself (generated via --build-id). * The build-id of the payload it depends on (extracted from the the previous payload or hypervisor during build time).

This means that the very first payload depends on the hypervisor build-id.

Not Yet Done

This is for further development of live patching.

TODO Goals

The implementation must also have a mechanism for (in no particular order):

Handle inlined LINE

This problem is related to hotpatch construction and potentially has influence on the design of the hotpatching infrastructure in Xen.

For example:

We have file1.c with functions f1 and f2 (in that order). f2 contains a BUG() (or WARN()) macro and at that point embeds the source line number into the generated code for f2.

Now we want to hotpatch f1 and the hotpatch source-code patch adds 2 lines to f1 and as a consequence shifts out f2 by two lines. The newly constructed file1.o will now contain differences in both binary functions f1 (because we actually changed it with the applied patch) and f2 (because the contained BUG macro embeds the new line number).

Without additional information, an algorithm comparing file1.o before and after hotpatch application will determine both functions to be changed and will have to include both into the binary hotpatch.

Options:

  1. Transform source code patches for hotpatches to be line-neutral for each chunk. This can be done in almost all cases with either reformatting of the source code or by introducing artificial preprocessor "#line n" directives to adjust for the introduced differences.

    This approach is low-tech and simple. Potentially generated backtraces and existing debug information refers to the original build and does not reflect hotpatching state except for actually hotpatched functions but should be mostly correct.

  2. Ignoring the problem and living with artificially large hotpatches that unnecessarily patch many functions.

    This approach might lead to some very large hotpatches depending on content of specific source file. It may also trigger pulling in functions into the hotpatch that cannot reasonable be hotpatched due to limitations of a hotpatching framework (init-sections, parts of the hotpatching framework itself, ...) and may thereby prevent us from patching a specific problem.

    The decision between 1. and 2. can be made on a patch--by-patch basis.

  3. Introducing an indirection table for storing line numbers and treating that specially for binary diffing. Linux may follow this approach.

    We might either use this indirection table for runtime use and patch that with each hotpatch (similarly to exception tables) or we might purely use it when building hotpatches to ignore functions that only differ at exactly the location where a line-number is embedded.

For BUG(), WARN(), etc., the line number is embedded into the bug frame, not the function itself.

Similar considerations are true to a lesser extent for FILE, but it could be argued that file renaming should be done outside of hotpatches.

Signature checking requirements.

The signature checking requires that the layout of the data in memory MUST be same for signature to be verified. This means that the payload data layout in ELF format MUST match what the hypervisor would be expecting such that it can properly do signature verification.

The signature is based on the all of the payloads continuously laid out in memory. The signature is to be appended at the end of the ELF payload prefixed with the string '~Module signature appended~\n', followed by an signature header then followed by the signature, key identifier, and signers name.

Specifically the signature header would be:

#define PKEY_ALGO_DSA       0  
#define PKEY_ALGO_RSA       1  

#define PKEY_ID_PGP         0 /* OpenPGP generated key ID */  
#define PKEY_ID_X509        1 /* X.509 arbitrary subjectKeyIdentifier */  

#define HASH_ALGO_MD4          0  
#define HASH_ALGO_MD5          1  
#define HASH_ALGO_SHA1         2  
#define HASH_ALGO_RIPE_MD_160  3  
#define HASH_ALGO_SHA256       4  
#define HASH_ALGO_SHA384       5  
#define HASH_ALGO_SHA512       6  
#define HASH_ALGO_SHA224       7  
#define HASH_ALGO_RIPE_MD_128  8  
#define HASH_ALGO_RIPE_MD_256  9  
#define HASH_ALGO_RIPE_MD_320 10  
#define HASH_ALGO_WP_256      11  
#define HASH_ALGO_WP_384      12  
#define HASH_ALGO_WP_512      13  
#define HASH_ALGO_TGR_128     14  
#define HASH_ALGO_TGR_160     15  
#define HASH_ALGO_TGR_192     16  


struct elf_payload_signature {  
    u8  algo;       /* Public-key crypto algorithm PKEY_ALGO_*. */  
    u8  hash;       /* Digest algorithm: HASH_ALGO_*. */  
    u8  id_type;    /* Key identifier type PKEY_ID*. */  
    u8  signer_len; /* Length of signer's name */  
    u8  key_id_len; /* Length of key identifier */  
    u8  __pad[3];  
    __be32  sig_len;    /* Length of signature data */  
};

(Note that this has been borrowed from Linux module signature code.).

.bss and .data sections.

In place patching writable data is not suitable as it is unclear what should be done depending on the current state of data. As such it should not be attempted.

That said we should provide hook functions so that the existing data can be changed during payload application.

To guarantee safety we disallow re-applying an payload after it has been reverted. This is because we cannot guarantee that the state of .bss and .data to be exactly as it was during loading. Hence the administrator MUST unload the payload and upload it again to apply it.

There is an exception to this: if the payload only has .livepatch.funcs; and the .data or .bss sections are of zero length.

Inline patching

The hypervisor should verify that the in-place patching would fit within the code or data.

Trampoline (e9 opcode), x86

The e9 opcode used for jmpq uses a 32-bit signed displacement. That means we are limited to up to 2GB of virtual address to place the new code from the old code. That should not be a problem since Xen hypervisor has a very small footprint.

However if we need - we can always add two trampolines. One at the 2GB limit that calls the next trampoline.

Please note there is a small limitation for trampolines in function entries: The target function (+ trailing padding) must be able to accomodate the trampoline. On x86 with +-2 GB relative jumps, this means 5 bytes are required which means that old_size MUST be at least five bytes if patching in trampoline.

Depending on compiler settings, there are several functions in Xen that are smaller (without inter-function padding).

 
readelf -sW xen-syms | grep " FUNC " | \
    awk '{ if ($3 < 5) print $3, $4, $5, $8 }'

...
3 FUNC LOCAL wbinvd_ipi
3 FUNC LOCAL shadow_l1_index
...

A compile-time check for, e.g., a minimum alignment of functions or a runtime check that verifies symbol size (+ padding to next symbols) for that in the hypervisor is advised.

The tool for generating payloads currently does perform a compile-time check to ensure that the function to be replaced is large enough.

Trampoline, ARM

The unconditional branch instruction (for the encoding see the DDI 0406C.c and DDI 0487A.j Architecture Reference Manual's). with proper offset is used for an unconditional branch to the new code. This means that that old_size MUST be at least four bytes if patching in trampoline.

The instruction offset is limited on ARM32 to +/- 32MB to displacement and on ARM64 to +/- 128MB displacement.

The new code is placed in the 8M - 10M virtual address space while the Xen code is in 2M - 4M. That gives us enough space.

The hypervisor also checks the displacement during loading of the payload.