Pith. sign in

REVIEW 4 major objections 5 minor 73 references

Exploiting Inaccurate Branch History in Side-Channel Attacks

T0 review · 4 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Branch predictors on modern CPUs can be forced to mis-speculate through their own history-update filters, yielding three Spectre-class attacks and a 24,628 bit/s kernel-memory leak with standard mitigations in place.

desk verdict Genuinely new ARM branch-predictor attacks with strong evidence; the Chimera and x86 BHS claims overreach and need to be reined in. read the letter →

arxiv 2506.07263 v1 pith:776FZ7HW submitted 2025-06-08 cs.CR cs.AR

Yuhui Zhu , Alessandro Biondi This is my paper
classification cs.CRcs.AR
keywords branchpredictionunithistorybufferspeculativeexecutionSpectresidechanneleBPFkernelmemorydisclosurebias-freepredictor
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper sets out to show that two little-documented features of modern branch predictors—bias-free branch prediction, which filters consistently behaving branches out of branch history, and branch history speculation, which records branch outcomes before they are confirmed—are exploitable security weaknesses, not just performance optimizations. It argues that an attacker who cannot directly inject branch history can still corrupt the Branch History Buffer indirectly, by evicting the records that decide whether a branch's footprint is written into history, or by letting speculated branches update history before they are resolved. The result is a family of three attacks that work across privilege boundaries with standard Spectre mitigations enabled: a control-flow side channel (BiasScope), a Spectre variant based on branch-status eviction (Spectre-BSE), and a variant based on branch-history speculation (Spectre-BHS). The paper closes with an eBPF-based demonstrator, Chimera, that leaks kernel memory at 24,628 bit/s on a Cortex-A76, which would matter because current mitigations were designed to stop direct branch-history injection, not indirect corruption of the update mechanism.

What carries the argument

The load-bearing machinery is the branch history's update logic rather than the prediction entries themselves. The paper treats the Branch History Buffer and Path History Register as a shift register of recent branch outcomes used to index prediction tables, and two hidden update filters as the attack surface: a Branch Status Table, a tagged, fully associative table that records whether each branch is biased and decides whether its footprint is written into history, and Branch History Speculation, which writes predicted outcomes into history before the branches resolve. The fallback mechanism of a TAGE (tagged geometric history length) predictor completes the machinery: once a shuffled, never-seen history is injected, later branches are predicted by program counter alone, allowing the attacker to splice together fragments of separate legitimate flows into one speculative path.

What would settle it

Run the paper's Spectre-BSE experiment on a Cortex-A72 in the cross-privilege setup, but with the evicting branch moved to a different process that shares only the same [15:4] address bits with the victim branch and with the standard BPU flush left enabled. If the cache signal still appears near the reported 99.9% success rate, the globally shared BST and cross-privilege claim is confirmed; if the signal disappears, the eviction effect requires the same address space or an unsanitized context switch and the claimed attack surface is narrower.

Watch

Extended reading notes

Core claim

The paper's central discovery is that the BHB update policy itself is attackable. On the Cortex-A72, a bias-free predictor maintains a Branch Status Table that filters out biased branches so their footprints never enter the path history; evicting a BST entry makes the next execution of that branch count as biased, so its footprint is silently dropped, which changes the BHB to match a different training flow and aliases the victim indirect branch's BTB entry. On the Cortex-A76 and A78AE, the BHB is instead updated speculatively from predicted, not yet resolved branches, and BTB/PHT eviction makes a taken branch be treated as not taken within the speculation window, again producing a false history. Both mechanisms let an unprivileged caller alter the branch history seen by a kernel-side indirect branch or by a later load-guarding conditional branch, causing speculative execution of a disclosure gadget despite existing BHI mitigations. The authors demonstrate the resulting leakage through a JIT-verified eBPF program that passes the verifier because the malicious path never exists architecturally.

Load-bearing premise

The attacks rest on the authors' interpretation of timing measurements as evidence of undocumented branch-predictor hardware—one global table shared by all contexts on a core, and branch history updating before branches are confirmed—and if either interpretation is wrong, the cross-privilege attacks would not generalize beyond the tested cores.

Editorial extensions

If this is right

  • If the paper's account is correct, ARM's current mitigation that populates the BHB on privilege transitions still leaves a gap, because the attack changes the update policy rather than the stored history values.
  • If correct, eviction-based mis-speculation on the Cortex-A76 and A78AE persists for a speculative window of more than 100 instructions, so real leakage gadgets with long prefix paths remain reachable.
  • If correct, conditional branches remain a cross-privilege target on x86 cores even when kernel indirect branches are protected by the recommended mitigations, because the corrupted history is produced before the conditional branch is predicted.
  • If correct, the eBPF verifier's architectural-path safety proof does not constrain speculative paths, and Chimera-like programs can reach secret-dependent loads at 24,628 bit/s on a Cortex-A76.
  • The attack surface is not uniform: the eviction vector did not reproduce on Intel cores, and the fallback-based Chimera vector did not reproduce on the Cortex-A72, so the vulnerable code patterns are microarchitecture-specific.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Beyond the paper: if the Branch Status Table is a single global structure shared by all contexts on the same core, then a sibling process or another execution context that can align a branch to the right address bits may perform the eviction without any syscall; the paper only demonstrated cross-process eviction when the standard BPU flush was disabled.
  • Beyond the paper: the 'not recorded counts as not taken' behavior for path-history registers suggests a finite-state model of branch-prediction records (taken, not-taken, not-recorded) that could be used to enumerate vulnerable branch sequences on other cores before writing any exploit code.
  • Beyond the paper: any JIT verifier that checks only architectural execution paths, not just the eBPF verifier, may admit Chimera-like programs; testing another verified in-kernel interpreter would show whether the issue is general.
  • Beyond the paper: a hardware fix would be to tag Branch Status Table entries with context information and to sanitize the update policy, not just the history bits, on privilege transitions; until then, newer context-tagging features may not close the attack.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. The paper investigates two underdocumented branch-predictor features—bias-free branch prediction (with a Branch Status Table, BST) and branch history speculation (BHS)—and claims three new attack primitives: Spectre-BSE, Spectre-BHS, and BiasScope, plus an eBPF-based demonstrator, Chimera, that leaks kernel memory at 24,628 bit/s on a Cortex-A76. The attacks are demonstrated empirically on several ARM cores, with partial results on AMD Zen4 and Intel cores. The central mechanism is that the attacker controls the presence of BST records through eviction or exploits speculated branch updates to the BHB/PHR to induce BHB aliasing and mis-speculation. The paper includes public artifacts and one CVE (CVE-2024-10929).

Significance. If the central claims hold, the paper makes a valuable contribution by exposing new BPU attack surfaces beyond previously known BHI attacks: BST eviction as a way to manipulate branch history without explicit branch injection, and BHS as a way to influence the BHB from transient branches. The empirical demonstrations on Cortex-A76/A78AE with high success rates, the kernel-space syscall-based BiasScope result, and the eBPF Chimera leak are concrete and reproducible (artifacts are provided). The paper is likely to interest the security and architecture communities. However, several headline claims in the abstract, contributions, and Table 3 exceed what the experimental sections actually demonstrate, so the paper needs a careful revision before it can be accepted.

major comments (4)
  1. [Abstract / §7.3 / Table 3] The abstract and the contribution bullet claim that the three attacks enable cross-privilege exploitation “with all Spectre mitigations enabled.” This is not consistent with the experiments: §7.3 states the Chimera eBPF demonstrator was tested in privileged mode because the verifier patch blocks unprivileged use, and that “this experiment was conducted with some mitigations disabled.” The headline leak rate of 24,628 bit/s is reported “under optimal conditions” and with mitigations disabled. Please state exactly which mitigations were disabled, whether the leak rate was measured with all recommended mitigations enabled in privileged mode, and clearly separate the demonstrated claim from the extrapolated one.
  2. [Table 3 and §8] Table 3 marks Spectre-BHS as exploitable on Zen4 and on Intel Gracemont/Redwood Cove/Crestmont via “M+C”/“E+C” for kernel conditional branches. Section 8, however, states “our efforts have not yielded positive results in this area” for x86 conditional-branch attacks, and §6.2 says the eviction-based variant could not be reproduced on Intel and was not reproduced on Zen4. These statements are in direct conflict. Either Table 3 must be corrected to “not exploitable” / “not demonstrated” for those rows, or §8 and §6.2 must be extended with the positive results. As published, the “demonstrated on multiple processors” claim for Spectre-BHS is not supported.
  3. [§5.4] Spectre-BSE is demonstrated only for in-place branch target training; the authors state “we currently limit our demonstration to in-place branch target training” and conjecture that a fully out-of-place target reuse attack is feasible. The abstract and contribution bullet present the three attacks as demonstrated without this caveat. Please either provide the out-of-place demonstration or qualify the claim, since the out-of-place vector is the one most relevant to the “naturally circumvent existing BHI mitigations” statement.
  4. [§5.2 / §5.3] The cross-process BiasScope result requires disabling the Spectre-v2 mitigation (BPU flush) for user-space processes; with mitigations enabled, the paper reports no mis-speculation signal across user-space processes. The demonstrated cross-privilege signal is the kernel-mode syscall experiment. The contribution bullet's phrasing “control flow monitoring across privilege boundaries with all Spectre mitigations enabled” should be narrowed to the syscall-based kernel leak, and the user-space cross-process limitation should be stated in the contributions.
minor comments (5)
  1. [§1] There is a typo in “Branch Hisotory Speculation”; it should read “Branch History Speculation.”
  2. [Table 3] The legend uses the garbled token “/adjus◎” in several cells; the intended meaning should be spelled out (e.g., “enabled for cross-privilege” or similar), since the current rendering is not self-explanatory.
  3. [§6.2, Figure 7] “Coretx-A76” should be “Cortex-A76.”
  4. [§5.2] The text says Bx_evict shares “at least 16 lower bits” with Bx_prime, while the BST is described as indexed by bits [15:4]; please clarify whether the requirement is that the lower 12 bits match or that bits [15:4] match.
  5. [Eq. (1)] The notation in Eq. (1) uses “BHB(BH[n]+Bi_probe)” but earlier text names the branch Bx_prime; please harmonize the notation for readability.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the attacks are empirical demonstrations with measured success rates; summary-vs-experiment inconsistencies are correctness concerns, not circular reasoning.

full rationale

The paper's derivation chain is empirical rather than definitional. The central claims—BST eviction on Cortex-A72, branch-history speculation across cores, the BHS fallback mechanism, and the Chimera leak rate—are supported by direct experiments (latency measurements, mis-speculation success rates, and end-to-end eBPF leakage) rather than by fitting a model to the claimed output. The 24,628 bit/s figure is a measured result under stated conditions, not a predicted value derived from fitted parameters. The BHB/BST attributions are hypotheses tested by constructing snippets and observing whether the predicted speculative behavior occurs; even if those attributions are wrong or incomplete, that is a correctness risk, not circularity. The paper compares against external prior work (BHI, BranchScope, ARM's Spectre-BHB documentation) and does not rely on a self-citation chain to ground its central premise. The inconsistency between the abstract's 'with all Spectre mitigations enabled' claim and Section 7.3's admission that 'this experiment was conducted with some mitigations disabled,' and the tension between Table 3's x86 BHS 'exploitable' entries and Section 8's 'our efforts have not yielded positive results in this area,' are overclaim/correctness issues that a reviewer should resolve, but they do not make any derivation equivalent to its inputs. No fitted parameter is renamed as a prediction, no cited uniqueness theorem is imported from the authors' own prior work, and no known result is merely relabeled. The paper is therefore not circular.

Assumptions & free parameters 0 free parameters · 3 assumptions · 0 invented entities

No new hardware entities are invented; the paper attributes observed behavior to undocumented existing features. The central claims rest on these behavioral attributions rather than on fitted free parameters.

assumptions (3)
  • domain assumption Cortex-A72 implements a Bias-Free Branch Predictor with a globally shared, tagged, full-associative Branch Status Table indexed by address bits [15:4].
    Inferred from behavioral experiments in Section 5.2, not from ARM documentation. The paper cites patents [2,17] for the concept. If the BST is partitioned or sanitized across contexts, the cross-privilege BiasScope and Spectre-BSE results may not generalize.
  • domain assumption Speculatively predicted branches update the BHB/PHR before architectural resolution (Branch History Speculation).
    Inferred from experiments in Section 6.2; the concept is supported by prior work [18,53]. The cascaded mis-speculation attack in Spectre-BHS depends on this update ordering.
  • domain assumption On unseen or shuffled branch history, the TAGE-based predictor falls back to the PC-based base predictor T0, and this fallback can be forced by a BHB-shuffle branch.
    Used for the Chimera attack in Section 7.3. The behavior is inferred from measurements and is not confirmed by hardware documentation.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Exploiting Inaccurate Branch History in Side-Channel Attacks." pith.science (2026). https://pith.science/paper/776FZ7HW

@misc{pith2026250607263,
  author       = {Pith},
  title        = {Pith review of: Exploiting Inaccurate Branch History in Side-Channel Attacks},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/776FZ7HW}},
  note         = {Machine review of arXiv:2506.07263}
}
read the original abstract

Modern out-of-order CPUs heavily rely on speculative execution for performance optimization, with branch prediction serving as a cornerstone to minimize stalls and maximize efficiency. Whenever shared branch prediction resources lack proper isolation and sanitization methods, they may originate security vulnerabilities that expose sensitive data across different software contexts. This paper examines the fundamental components of modern Branch Prediction Units (BPUs) and investigates how resource sharing and contention affect two widely implemented but underdocumented features: Bias-Free Branch Prediction and Branch History Speculation. Our analysis demonstrates that these BPU features, while designed to enhance speculative execution efficiency through more accurate branch histories, can also introduce significant security risks. We show that these features can inadvertently modify the Branch History Buffer (BHB) update behavior and create new primitives that trigger malicious mis-speculations. This discovery exposes previously unknown cross-privilege attack surfaces for Branch History Injection (BHI). Based on these findings, we present three novel attack primitives: two Spectre attacks, namely Spectre-BSE and Spectre-BHS, and a cross-privilege control flow side-channel attack called BiasScope. Our research identifies corresponding patterns of vulnerable control flows and demonstrates exploitation on multiple processors. Finally, Chimera is presented: an attack demonstrator based on eBPF for a variant of Spectre-BHS that is capable of leaking kernel memory contents at 24,628 bit/s.

Figures

Figures reproduced from arXiv: 2506.07263 by the authors.

Figure 1
Figure 1. TAGE branch predictor. outcomes on the execution path, typically as a shift register of taken/not-taken bits, which length is pre-defined. Some im￾plementations employ an enhanced variant of BHB known as the Path History Register (PHR) [36, 47]. Unlike the canon￾ical BHB, which only records taken/not-taken outcomes, the PHR maintains a complete jumping path by storing multi-bit footprints that encode both the source… view at source ↗
Figure 2
Figure 2. Cache access latency (ns) as a function of the num [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. BHB update process in bias-free branch prediction. [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: Monitoring a victim branch using BiasScope. [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: Error rate of BST side channels under intra-process [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]
Figure 6
Figure 6. Figure 6: BHB aliasing in Spectre-BSE. At this stage, the environment is almost prepared, and the attacker is ready to proceed with the malicious actions. The attacker performs a targeted BST eviction on Bev = BH[4], then invokes the vulnerable code snippet with FB. As depicted …
Figure 7
Figure 7. Figure 7: Access latency of data cache probe with varying [PITH_FULL_IMAGE:figures/full_fig_p011_7.png]
Figure 8
Figure 8. Figure 8: Transitions in BPU record status for a branch, in [PITH_FULL_IMAGE:figures/full_fig_p020_8.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

73 extracted references · 73 canonical work pages

  1. [1]

    https://docs.kernel.org/admin-guide/h w-vuln/spectre.html

    The Linux Kernel documentation - Spectre Side Chan- nels. https://docs.kernel.org/admin-guide/h w-vuln/spectre.html

    work page

  2. [2]

    Al-Otoom, Paul Caprioli, and Jeffrey J

    Muawya M. Al-Otoom, Paul Caprioli, and Jeffrey J. Cook. Detecting and Filtering Biased Branches in Global Branch History, June 2014. https://pate nts.google.com/patent/US20140156978A1/en

    work page 2014

  3. [3]

    Software Optimization Guide for the AMD Zen4 Microarchitecture, January 2023

    AMD. Software Optimization Guide for the AMD Zen4 Microarchitecture, January 2023

    work page 2023

  4. [4]

    Trusted Firmware-A 2.12.0 documentation: 9.6

    ARM. Trusted Firmware-A 2.12.0 documentation: 9.6. Advisory TFV-6 (CVE-2017-5753, CVE-2017-5715, CVE-2017-5754). https://trustedfirmware-a .readthedocs.io/en/latest/security_advisor ies/security-advisory-tfv-6.html , June 2018

    work page 2017

  5. [5]

    Straight-line Speculation Whitepaper

    ARM. Straight-line Speculation Whitepaper. https: //developer.arm.com/documentation/102825/l atest/, June 2020

    work page 2020

  6. [6]

    Spectre-BHB: Speculative Target Reuse Attacks version 1.8-r0p2

    ARM. Spectre-BHB: Speculative Target Reuse Attacks version 1.8-r0p2. https://developer.arm.com/do cumentation/102898/0107/?lang=en , December 2023

    work page 2023

  7. [7]

    Arm Architecture Reference Manual for A- profile architecture version L.a

    ARM. Arm Architecture Reference Manual for A- profile architecture version L.a. https://developer. arm.com/documentation/ddi0487/la/?lang=en , December 2024

    work page 2024

  8. [8]

    Branch history injection: On the effectiveness of hardware mitigations against Cross-Privilege spectre-v2 attacks

    Enrico Barberis, Pietro Frigo, Marius Muench, Herbert Bos, and Cristiano Giuffrida. Branch history injection: On the effectiveness of hardware mitigations against Cross-Privilege spectre-v2 attacks. In 31st USENIX Security Symposium (USENIX Security 22), pages 971– 988, Boston, MA, August 2022. USENIX Association. https://www.usenix.org/conference/usenixs...

    work page 2022

Show all 73 references
  1. [9]

    A Systematic Evaluation of Transient Execution Attacks and Defenses

    Claudio Canella, Jo Van Bulck, Michael Schwarz, Moritz Lipp, Benjamin von Berg, Philipp Ortner, Frank Piessens, Dmitry Evtyushkin, and Daniel Gruss. A Systematic Evaluation of Transient Execution Attacks and Defenses. In 28th USENIX Security Symposium (USENIX Security 19), pag...

    2019

  2. [10]

    Po-Yung Chang, Marius Evers, and Yale N. Patt. Im- proving branch prediction accuracy by reducing pattern history table interference. International Journal of Par- allel Programming, 25(5):339–362, October 1997

    1997

  3. [11]

    Po-Yung Chang, Eric Hao, and Yale N. Patt. Target pre- diction for indirect jumps. ACM SIGARCH Computer Architecture News, 25(2):274–283, May 1997

    1997

  4. [12]

    Re: [PATCH 0/3] arm64: Proton-pack: Add Spectre-BSE mitigation for Cortex-A7{2,3,5} - Doebel, Bjoern

    Bjoern Doebel. Re: [PATCH 0/3] arm64: Proton-pack: Add Spectre-BSE mitigation for Cortex-A7{2,3,5} - Doebel, Bjoern. https://lore.kernel.org/li nux-arm-kernel/965352b6-c21f-4161-9f23-2e9 6cc1267f6@amazon.de/, January 2025

    2025

  5. [13]

    Driesen and U

    K. Driesen and U. Holzle. The cascaded predictor: Eco- nomical and adaptive branch target prediction. In Pro- ceedings. 31st Annual ACM/IEEE International Sympo- sium on Microarchitecture, pages 249–258, December 1998

    1998

  6. [14]

    BranchScope: A New Side-Channel Attack on Directional Branch Pre- dictor

    Dmitry Evtyushkin, Ryan Riley, Nael CSE and ECE Abu-Ghazaleh, and Dmitry Ponomarev. BranchScope: A New Side-Channel Attack on Directional Branch Pre- dictor. ACM SIGPLAN Notices, 53(2):693–707, March 2018. 15

    2018

  7. [15]

    SQUIP: Exploiting the Sched- uler Queue Contention Side Channel

    Stefan Gast, Jonas Juffinger, Martin Schwarzl, Gururaj Saileshwar, Andreas Kogler, Simone Franza, Markus Köstl, and Daniel Gruss. SQUIP: Exploiting the Sched- uler Queue Contention Side Channel. In 2023 IEEE Symposium on Security and Privacy (SP), pages 2256– 2272, May 2023

    2023

  8. [16]

    Checkpoint allo- cation and release

    Amit Golander and Shlomo Weiss. Checkpoint allo- cation and release. ACM Trans. Archit. Code Optim., 6(3):10:1–10:27, October 2009

    2009

  9. [17]

    Dibakar Gope and Mikko H. Lipasti. Bias-Free Branch Predictor. In 2014 47th Annual IEEE/ACM Interna- tional Symposium on Microarchitecture, pages 521–532, December 2014

    2014

  10. [18]

    Eric Hao, Po-Yung Chang, and Yale N. Patt. The ef- fect of speculatively updating branch history on branch prediction accuracy, revisited. In Proceedings of the 27th Annual International Symposium on Microarchitec- ture, MICRO 27, pages 228–232, New York, NY , USA, November 19...

    1994

  11. [19]

    Speculation at fault: Modeling and testing microarchitectural leakage of CPU exceptions

    Jana Hofmann, Emanuele Vannacci, Cedric Fournet, Boris Kopf, and Oleksii Oleksenko. Speculation at fault: Modeling and testing microarchitectural leakage of CPU exceptions. In 32nd USENIX Security Sym- posium (USENIX Security 23), pages 7143–7160, Ana- heim, CA, August 2023. U...

    2023

  12. [20]

    Speculative Execution Side Channel Mitigations

    Intel. Speculative Execution Side Channel Mitigations. https://www.intel.com/content/www/us/en/de veloper/articles/technical/software-secur ity-guidance/technical-documentation/specu lative-execution-side-channel-mitigations .html

  13. [21]

    Indirect Branch Predictor Barrier

    Intel. Indirect Branch Predictor Barrier. https://www. intel.com/content/www/us/en/developer/arti cles/technical/software-security-guidance/ technical-documentation/indirect-branch-p redictor-barrier.html, March 2018

    2018

  14. [22]

    Overtake: Achieving Meltdown-type Attacks with One Instruction

    Yu Jin, Pengfei Qiu, Chunlu Wang, Yihao Yang, Dong- sheng Wang, Xiaoyong Li, Qian Wang, and Gang Qu. Overtake: Achieving Meltdown-type Attacks with One Instruction. In 2023 Asian Hardware Oriented Security and Trust Symposium (AsianHOST), pages 1–6, Decem- ber 2023

    2023

  15. [23]

    Kasper: Scanning for generalized transient execution gadgets in the linux kernel

    Brian Johannesmeyer, Jakob Koschel, Kaveh Razavi, Herbert Bos, and Cristiano Giuffrida. Kasper: Scanning for generalized transient execution gadgets in the linux kernel. In NDSS Symposium, volume 2022, 2022. http s://www.ndss-symposium.org/ndss-paper/aut o-draft-247/

    2022

  16. [24]

    D. R. Kaeli and P. G. Emma. Branch history table predic- tion of moving target branches due to subroutine returns. In 1991The 18th Annual International Symposium on Computer Architecture, pages 34–42. IEEE Computer Society, January 1991

    1991

  17. [25]

    The gates of time: Improving cache attacks with transient execution

    Daniel Katzman, William Kosasih, Chitchanok Chuengsatiansup, Eyal Ronen, and Yuval Yarom. The gates of time: Improving cache attacks with transient execution. In 32nd USENIX Security Symposium (USENIX Security 23), pages 1955–1972, Anaheim, CA, August 2023. USENIX Association....

    1955

  18. [26]

    An Analysis of Speculative Type Confusion Vulnerabilities in the Wild

    Ofek Kirzner and Adam Morrison. An Analysis of Speculative Type Confusion Vulnerabilities in the Wild. In 30th USENIX Security Symposium (USENIX Security 21), pages 2399–2416, 2021. https://www.usenix.o rg/conference/usenixsecurity21/presentatio n/kirzner

    2021

  19. [27]

    The bimode++ branch predictor

    Kenji Kise, Takahiro Katagiri, Hiroki Honda, and Toshit- sugu Yuba. The bimode++ branch predictor. In Innovative Architecture for Future Generation High- Performance Processors and Systems (IWIA’05), pages 8 pp.–, January 2005

    2005

  20. [28]

    Spectre Attacks: Exploiting Speculative Execution

    Paul Kocher, Jann Horn, Anders Fogh, Daniel Genkin, Daniel Gruss, Werner Haas, Mike Hamburg, Moritz Lipp, Stefan Mangard, Thomas Prescher, Michael Schwarz, and Yuval Yarom. Spectre Attacks: Exploiting Speculative Execution. In 2019 IEEE Symposium on Security and Privacy (SP), ...

    2019

  21. [29]

    Kha- sawneh, Chengyu Song, and Nael Abu-Ghazaleh

    Esmaeil Mohammadian Koruyeh, Khaled N. Kha- sawneh, Chengyu Song, and Nael Abu-Ghazaleh. Spec- tre Returns! Speculation Attacks using the Return Stack Buffer. In 12th USENIX Workshop on Offensive Tech- nologies (WOOT 18), 2018. https://www.usenix.o rg/conference/woot18/present...

    2018

  22. [30]

    D. J. Lalja. Reducing the branch penalty in pipelined processors. Computer, 21(7):47–55, July 1988

    1988

  23. [31]

    Indi- rector: High-Precision branch target injection attacks exploiting the indirect branch predictor

    Luyi Li, Hosein Yavarzadeh, and Dean Tullsen. Indi- rector: High-Precision branch target injection attacks exploiting the indirect branch predictor. In33rd USENIX Security Symposium (USENIX Security 24), pages 2137– 2154, Philadelphia, PA, August 2024. USENIX Associ- ation. ht...

    2024

  24. [32]

    Ret2spec: Speculative Execution Using Return Stack Buffers

    Giorgi Maisuradze and Christian Rossow. Ret2spec: Speculative Execution Using Return Stack Buffers. In Proceedings of the 2018 ACM SIGSAC Conference on Computer and Communications Security, CCS ’18, 16 pages 2109–2122, New York, NY , USA, October 2018. Association for Computin...

    2018

  25. [33]

    Bypassing memory safety mechanisms through speculative control flow hijacks

    Andrea Mambretti, Alexandra Sandulescu, Alessan- dro Sorniotti, William Robertson, Engin Kirda, and Anil Kurmus. Bypassing memory safety mechanisms through speculative control flow hijacks. In 2021 IEEE European Symposium on Security and Privacy (Eu- roS&P), pages 633–649, Sep...

    2021

  26. [34]

    SpectrEM: Exploiting electromagnetic emanations dur- ing transient execution

    Jesse De Meulemeester, Antoon Purnal, Lennert Wouters, Arthur Beckers, and Ingrid Verbauwhede. SpectrEM: Exploiting electromagnetic emanations dur- ing transient execution. In 32nd USENIX Security Sym- posium (USENIX Security 23), pages 6293–6310, Ana- heim, CA, August 2023. U...

    2023

  27. [35]

    You Cannot Always Win the Race: Analyzing mitigations for branch target prediction attacks

    Alyssa Milburn, Ke Sun, and Henrique Kawakami. You Cannot Always Win the Race: Analyzing mitigations for branch target prediction attacks. In 2023 IEEE 8th Eu- ropean Symposium on Security and Privacy (EuroS&P), pages 671–686, July 2023

    2023

  28. [36]

    Dynamic path-based branch correlation

    Ravi Nair. Dynamic path-based branch correlation. In Proceedings of the 28th Annual International Sym- posium on Microarchitecture, pages 15–23, November 1995

    1995

  29. [37]

    Hide and Seek with Spectres: Effi- cient discovery of speculative information leaks with random testing

    Oleksii Oleksenko, Marco Guarnieri, Boris Köpf, and Mark Silberstein. Hide and Seek with Spectres: Effi- cient discovery of speculative information leaks with random testing. In 2023 IEEE Symposium on Security and Privacy (SP), pages 1737–1752, May 2023

    2023

  30. [38]

    ShowTime: Amplifying Arbitrary CPU Timing Side Channels

    Antoon Purnal, Marton Bognar, Frank Piessens, and In- grid Verbauwhede. ShowTime: Amplifying Arbitrary CPU Timing Side Channels. In Proceedings of the 2023 ACM Asia Conference on Computer and Communica- tions Security, ASIA CCS ’23, pages 205–217, New York, NY , USA, July 2023...

    2023

  31. [39]

    SpecTaint: Speculative taint analysis for discovering spectre gadgets

    Zhenxiao Qi, Qian Feng, Yueqiang Cheng, Mengjia Yan, Peng Li, Heng Yin, and Tao Wei. SpecTaint: Speculative taint analysis for discovering spectre gadgets. In NDSS Symposium, 2021

    2021

  32. [40]

    PMU-Spill: Performance Monitor Unit Coun- ters Leak Secrets in Transient Executions

    Pengfei Qiu, Qiang Gao, Dongsheng Wang, Yongqiang Lyu, Chang Liu, Xiaoyong Li, Chunlu Wang, and Gang Qu. PMU-Spill: Performance Monitor Unit Coun- ters Leak Secrets in Transient Executions. In 2022 Asian Hardware Oriented Security and Trust Sympo- sium (AsianHOST), pages 1–6, ...

    2022

  33. [41]

    GhostRace: Exploiting and mitigating speculative race conditions

    Hany Ragab, Andrea Mambretti, Anil Kurmus, and Cris- tiano Giuffrida. GhostRace: Exploiting and mitigating speculative race conditions. In 33rd USENIX Security Symposium (USENIX Security 24), pages 6185–6202, Philadelphia, PA, August 2024. USENIX Association. https://www.useni...

    2024

  34. [42]

    PACMAN: Attacking ARM pointer au- thentication with speculative execution

    Joseph Ravichandran, Weon Taek Na, Jay Lang, and Mengjia Yan. PACMAN: Attacking ARM pointer au- thentication with speculative execution. In Proceed- ings of the 49th Annual International Symposium on Computer Architecture, ISCA ’22, pages 685–698, New York, NY , USA, June 2022...

    2022

  35. [43]

    Tullsen, and Ashish Venkat

    Xida Ren, Logan Moody, Mohammadkazem Taram, Matthew Jordan, Dean M. Tullsen, and Ashish Venkat. I See Dead Mops: Leaking Secrets via Intel/AMD Micro- Op Caches. In 2021 ACM/IEEE 48th Annual Interna- tional Symposium on Computer Architecture (ISCA) , pages 361–374, June 2021

    2021

  36. [44]

    Reverse Engineering the Stream Prefetcher for Profit

    Aditya Rohan, Biswabandan Panda, and Prakhar Agar- wal. Reverse Engineering the Stream Prefetcher for Profit. In 2020 IEEE European Symposium on Security and Privacy Workshops (EuroS&PW), pages 682–687, September 2020

    2020

  37. [45]

    FetchBench: Systematic Identification and Characteri- zation of Proprietary Prefetchers

    Till Schlüter, Amit Choudhari, Lorenz Hetterich, Leon Trampert, Hamed Nemati, Ahmad Ibrahim, Michael Schwarz, Christian Rossow, and Nils Ole Tippenhauer. FetchBench: Systematic Identification and Characteri- zation of Proprietary Prefetchers. In Proceedings of the 2023 ACM SIG...

    2023

  38. [46]

    Linux Socket Filtering aka Berkeley Packet Filter (BPF)

    Jay Schulist, Daniel Borkmann, and Alexei Starovoitov. Linux Socket Filtering aka Berkeley Packet Filter (BPF). https://www.kernel.org/doc/Documentation/n etworking/filter.txt, 2024

    2024

  39. [47]

    A. Seznec. Analysis of the O-GEometric history length branch predictor. In 32nd International Symposium on Computer Architecture (ISCA’05), pages 394–405, June 2005

    2005

  40. [48]

    A 64 kbytes ISL-TAGE branch predictor

    André Seznec. A 64 kbytes ISL-TAGE branch predictor. In JWAC-2: Championship Branch Prediction, 2011

    2011

  41. [49]

    A 64-Kbytes ITTAGE indirect branch predictor

    André Seznec. A 64-Kbytes ITTAGE indirect branch predictor. In JWAC-2: Championship Branch Prediction, 2011. 17

    2011

  42. [50]

    A new case for the TAGE branch pre- dictor

    André Seznec. A new case for the TAGE branch pre- dictor. In 2011 44th Annual IEEE/ACM International Symposium on Microarchitecture (MICRO), pages 117– 127, December 2011. https://ieeexplore.ieee.or g/abstract/document/7851464

    2011

  43. [51]

    A case for (partially) tagged geometric history length branch prediction

    André Seznec and Pierre Michaud. A case for (partially) tagged geometric history length branch prediction. Jour- nal of Instruction-Level Parallelism, (8):1–23, 2006

    2006

  44. [52]

    Spectre Declassified: Reading from the Right Place at the Wrong Time

    Basavesh Ammanaghatta Shivakumar, Jack Barnes, Gilles Barthe, Sunjay Cauligi, Chitchanok Chuengsa- tiansup, Daniel Genkin, Sioli O’Connell, Peter Schwabe, Rui Qi Sim, and Yuval Yarom. Spectre Declassified: Reading from the Right Place at the Wrong Time. In 2023 IEEE Symposium ...

    2023

  45. [53]

    Speculative updates of local and global branch history: A quantitative analysis

    Kevin Skadron, Margaret Martonosi, and Douglas Clark. Speculative updates of local and global branch history: A quantitative analysis. Journal of Instruction-Level Parallelism, 2, 2000. http://www.jilp.org/vol2/ v2paper1.pdf

    2000

  46. [54]

    Chappell, Mitch Alsup, and Yale N

    Eric Sprangle, Robert S. Chappell, Mitch Alsup, and Yale N. Patt. The agree predictor: A mechanism for re- ducing negative branch history interference. SIGARCH Comput. Archit. News, 25(2):284–291, May 1997

    1997

  47. [55]

    Invisible Probe: Timing Attacks with PCIe Congestion Side-channel

    Mingtian Tan, Junpeng Wan, Zhe Zhou, and Zhou Li. Invisible Probe: Timing Attacks with PCIe Congestion Side-channel. In 2021 IEEE Symposium on Security and Privacy (SP), pages 322–338, May 2021

    2021

  48. [56]

    TLB;DR: Enhancing TLB-based attacks with TLB desynchronized reverse engineering

    Andrei Tatar, Daniël Trujillo, Cristiano Giuffrida, and Herbert Bos. TLB;DR: Enhancing TLB-based attacks with TLB desynchronized reverse engineering. In 31st USENIX Security Symposium (USENIX Security 22) , pages 989–1007, Boston, MA, August 2022. USENIX Association. https://w...

    2022

  49. [57]

    Inception: Exposing new attack surfaces with train- ing in transient execution

    Daniël Trujillo, Johannes Wikner, and Kaveh Razavi. Inception: Exposing new attack surfaces with train- ing in transient execution. In 32nd USENIX Secu- rity Symposium (USENIX Security 23) , pages 7303– 7320, Anaheim, CA, August 2023. USENIX Associ- ation. https://www.usenix.o...

    2023

  50. [58]

    Fletcher, and David Kohlbrenner

    Jose Rodrigo Sanchez Vicarte, Michael Flanders, Ric- cardo Paccagnella, Grant Garrett-Grossman, Adam Mor- rison, Christopher W. Fletcher, and David Kohlbrenner. Augury: Using Data Memory-Dependent Prefetchers to Leak Data at Rest. In2022 IEEE Symposium on Security and Privacy ...

    2022

  51. [59]

    InSpectre gadget: Inspecting the residual attack surface of cross-privilege spectre v2

    Sander Wiebing, Alvise de Faveri Tron, Herbert Bos, and Cristiano Giuffrida. InSpectre gadget: Inspecting the residual attack surface of cross-privilege spectre v2. In 33rd USENIX Security Symposium (USENIX Secu- rity 24), pages 577–594, Philadelphia, PA, August 2024. USENIX A...

    2024

  52. [60]

    Training Solo: On the Limitations of Domain Isolation Against Spectre- v2 Attacks

    Sander Wiebing and Cristiano Giuffrida. Training Solo: On the Limitations of Domain Isolation Against Spectre- v2 Attacks. In 2025 IEEE Symposium on Security and Privacy (SP), pages 3599–3616. IEEE Computer Soci- ety, May 2025

    2025

  53. [61]

    The AMD Branch (Mis)predictor Part 2: Where No CPU has Gone Before (CVE-2021-26341)

    Pawel Wieczorkiewicz. The AMD Branch (Mis)predictor Part 2: Where No CPU has Gone Before (CVE-2021-26341). h t t p s : //grsecurity.net/amd_branch_mispredict or_part_2_where_no_cpu_has_gone_before , March 2022

    2021

  54. [62]

    RETBLEED: Arbi- trary speculative code execution with return instructions

    Johannes Wikner and Kaveh Razavi. RETBLEED: Arbi- trary speculative code execution with return instructions. In 31st USENIX Security Symposium (USENIX Secu- rity 22), pages 3825–3842, Boston, MA, August 2022. USENIX Association. https://www.usenix.org/c onference/usenixsecurit...

    2022

  55. [63]

    Breaking the Bar- rier: Post-Barrier Spectre Attacks

    Johannes Wikner and Kaveh Razavi. Breaking the Bar- rier: Post-Barrier Spectre Attacks. In 2025 IEEE Sympo- sium on Security and Privacy (SP), pages 89–89. IEEE Computer Society, November 2024

    2025

  56. [64]

    Phantom: Exploiting Decoder-detectable Mispredic- tions

    Johannes Wikner, Daniël Trujillo, and Kaveh Razavi. Phantom: Exploiting Decoder-detectable Mispredic- tions. In Proceedings of the 56th Annual IEEE/ACM International Symposium on Microarchitecture, MICRO ’23, pages 49–61, New York, NY , USA, December 2023. Association for Comp...

    2023

  57. [65]

    Rendering contention channel made practical in web browsers

    Shujiang Wu, Jianjia Yu, Min Yang, and Yinzhi Cao. Rendering contention channel made practical in web browsers. In 31st USENIX Security Symposium (USENIX Security 22), pages 3183–3199, Boston, MA, August 2022. USENIX Association. https://www.us enix.org/conference/usenixsecuri...

    2022

  58. [66]

    Pathfinder: High-Resolution Control-Flow At- tacks Exploiting the Conditional Branch Predictor

    Hosein Yavarzadeh, Archit Agarwal, Max Christman, Christina Garman, Daniel Genkin, Andrew Kwong, Daniel Moghimi, Deian Stefan, Kazem Taram, and Dean Tullsen. Pathfinder: High-Resolution Control-Flow At- tacks Exploiting the Conditional Branch Predictor. In Proceedings of the 2...

    2024

  59. [67]

    Half&Half: Demystifying Intel’s Directional Branch Predictors for Fast, Secure Partitioned Execution

    Hosein Yavarzadeh, Mohammadkazem Taram, Shravan Narayan, Deian Stefan, and Dean Tullsen. Half&Half: Demystifying Intel’s Directional Branch Predictors for Fast, Secure Partitioned Execution. In 2023 IEEE Sym- posium on Security and Privacy (SP), pages 1220–1237, May 2023

    2023

  60. [68]

    Tse-Yu Yeh and Yale N. Patt. Two-level adaptive train- ing branch prediction. InProceedings of the 24th Annual International Symposium on Microarchitecture, MICRO 24, pages 51–61, New York, NY , USA, September 1991. Association for Computing Machinery

    1991

  61. [69]

    Tse-Yu Yeh and Yale N. Patt. Alternative implementa- tions of two-level adaptive branch prediction.SIGARCH Comput. Archit. News, 20(2):124–134, April 1992

    1992

  62. [70]

    Fletcher

    Jiyong Yu, Aishani Dutta, Trent Jaeger, David Kohlbren- ner, and Christopher W. Fletcher. Synchronization stor- age channels (S2C): Timer-less cache Side-Channel attacks on the apple M1 via hardware synchroniza- tion instructions. In 32nd USENIX Security Sympo- sium (USENIX Se...

    1973

  63. [71]

    (M)WAIT for it: Bridging the gap between microarchitectural and architectural side channels

    Ruiyi Zhang, Taehyun Kim, Daniel Weber, and Michael Schwarz. (M)WAIT for it: Bridging the gap between microarchitectural and architectural side channels. In 32nd USENIX Security Symposium (USENIX Security 23), pages 7267–7284, Anaheim, CA, August 2023. USENIX Association. http...

    2023

  64. [72]

    Exploring Branch Predictors for Constructing Transient Execution Trojans

    Tao Zhang, Kenneth Koltermann, and Dmitry Ev- tyushkin. Exploring Branch Predictors for Constructing Transient Execution Trojans. In Proceedings of the Twenty-Fifth International Conference on Architectural Support for Programming Languages and Operating Systems, ASPLOS ’20, p...

    2020

  65. [73]

    not recorded

    Zhiyuan Zhang, Mingtian Tao, Sioli O’Connell, Chitchanok Chuengsatiansup, Daniel Genkin, and Yuval Yarom. BunnyHop: Exploiting the instruction prefetcher. In 32nd USENIX Security Symposium (USENIX Secu- rity 23), pages 7321–7337, Anaheim, CA, August 2023. USENIX Association. h...

    2023

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.